2011 special edition - converged infrastructure
DESCRIPTION
TRANSCRIPT
your strategic guide to
converged
From the editors oF
Networks key to convergence puzzle
A hands-on guide to convergence
how to staff for converged it
managing data center sprawlin
sid
e
infrastructure
03IDGEconv_cover.indd 1 8/24/11 2:03:08 PM
2 C o n v e r g e d I n f r a s t r u C t u r e
h i s i s a t i m e o f trans forma tion for IT. The old method of dedicating infrastructure to individual projects has proven unsustainable, resulting in both underutilized
capacity and too much equipment to manage. Converged infrastructure seeks to consolidate compute, storage, and network resources into a shared fabric that scales easily, supports reconfiguration on the fly, and extracts maximum utility from the hardware at hand. This special guide explores how companies are moving to converged infrastructure and the benefits they expect to gain. Along the way, you’ll find explanations of the key technical and management issues that arise in the transition to a modern, agile data center architecture.
Tbuilding the modern data center
Your strategic Guide to Converged infrastructure
Brought to you by the editors of CIO, Computerworld, CSO, InfoWorld, ITworld and Network World
E d i t o r i a l
Editor in Chief, infoWorlderic Knorr
Executive Editor, ComputerworldMitch Betts
Editor in Chief, Network WorldJohn dix
Editor in Chief, Cio.comBrian Carlson
Managing Editor, ComputerworldBob rawson
Managing Editor, Network Worldryan francis
Copy EditorsMelissa andersenColleen Barry
Executive director, art and designMary Lester
i d G E N t E r p r i s E492 old Connecticut PathP.o. Box 9208framingham, Ma 01701-9208(508) 879-0700
E M a i l first initial last name (no space) @idgenterprise.com
© Idg Communications Inc. 2011
iNtErNatioNal data GroupChairman of the BoardPatrick J. McgovernCEo, idG CommunicationsBob Carrigan
Idg enterprise is a business unit of Idg, the world’s leading technology media, research and events company. Idg publishes more than 300 magazines and newspapers and offers online users the largest network of technology-specific sites around the world through Idg.net (www.idg.net), which comprises more than 330 targeted Web sites in 80 countries. Idg is also a leading producer of 168 computer-related events worldwide, and Idg’s research company, IdC, provides global market intelligence and advice through 51 offices in 43 countries. Company information is available at www.idg.com.
From the editors oF
03IDGEconv_0Edit Letter, TOC.indd 2 8/24/11 1:49:54 PM
C o n v e r g e d I n f r a s t r u C t u r e 3
managing data center sprawllearn from three companies that are reaping the benefits of deploying a simpler infrastructure B y B o B V i o l i N o
Networks key to convergence puzzleupgrading the network helps get the most out of server infrastructure, simplifies wiring, and paves the way for the cloud B y M a r i a Ko r o loV
a hands-on guide to convergencea typical midsize data center gets a makeover — providing a detailed example of how to deploy a modern, converged infrastructure B y M at t p r i G G E
how to staff for converged itas convergence shakes up the data center, finding the right people who are highly versatile and cross-disciplined is key for Cios who want to stay ahead of the curve B y pa M B a K E r
6
14
22
30
inside
03IDGEconv_0Edit Letter, TOC.indd 3 8/24/11 1:49:57 PM
www.f5.com
Missing the agility from your virtual infrastructure?
Find it with optimized application delivery.Read the Nemertes Research paper to see why
optimized application delivery is critical to breaking
through network limitations and realizing the highest
levels of performance and reliability from your virtual
environment.
F5 solutions for optimizing application delivery include:
• Application awareness
• Virtual platform integration
• Service offloading
• Application availability and monitoring
IT agility. Your way.
C
M
Y
CM
MY
CY
CMY
K
F5_Ad_2_11_FinalART.pdf 1 2/11/11 2:53 PM
www.f5.com
Missing the agility from your virtual infrastructure?
Find it with optimized application delivery.Read the Nemertes Research paper to see why
optimized application delivery is critical to breaking
through network limitations and realizing the highest
levels of performance and reliability from your virtual
environment.
F5 solutions for optimizing application delivery include:
• Application awareness
• Virtual platform integration
• Service offloading
• Application availability and monitoring
IT agility. Your way.
C
M
Y
CM
MY
CY
CMY
K
F5_Ad_2_11_FinalART.pdf 1 2/11/11 2:53 PM
managing data center sprawl
Learn from three companies that are reaping
the benefits of deploying a simpler infrastructure
By BoB VioLino
6 C o n v e r g e d I n f r a s t r u C t u r e | d ata c e n t e r m a n a g e m e n t
03IDGEconv_1data.indd 6 8/24/11 1:53:05 PM
oncerned about
the cost and complexity of managing a hodge-podge of equipment in
your data center?Some IT shops are responding
by deploying a converged infra-structure, in which multiple IT components, namely servers, storage systems, networking equipment and IT infrastructure management software, are pack-aged together in a single offering.
C 7
03IDGEconv_1data.indd 7 8/24/11 1:53:06 PM
8 C o n v e r g e d I n f r a s t r u C t u r e | d ata c e n t e r m a n a g e m e n t
A converged infrastructure lets organizations centralize the management of their IT resources in hopes of achieving benefits such as systems consolidation, improved resource utilization rates and lower costs.
The concept of a converged infrastructure emerged because many data centers have gotten out of control in recent years with the explosion of data and proliferation of servers. The rise in processing and storage requirements has led to higher energy consumption and increased maintenance and opera-tions costs in many data centers.
The converged infrastructure
addresses the issues of server sprawl and archi-tectural silos of traditional data centers by pooling IT resources so they can be
shared by multiple applications and lines of business. More-over, with the ability to pool IT resources, automate provision-ing and ratchet up capacity, a converged infrastructure can also set the stage for a private cloud.
Organizations in a variety of industries have launched
converged infrastructure strategies and are seeing tangible benefits.
Keeping up with business growthWalz Group, a Temecula, Calif., firm that handles critical documents for customers, has seen dramatic growth in its data storage require-ments in recent years. The firm’s customers, including some of the largest financial institutions, government agencies, healthcare providers and insurance companies in the U.S., have stringent needs for compliance-related document management and archiving.
Walz Group generates and
“ We’ve seen huge operational efficiency improvements, greater ability to scale, security improvements … and improved disaster recovery.”
B a r t fa l z a r a n o, C h I e f I n f o r m at I o n s e C u r I t y o f f I C e r , Wa l z g r o u p
03IDGEconv_1data.indd 8 8/24/11 1:53:06 PM
d ata c e n t e r m a n a g e m e n t | C o n v e r g e d I n f r a s t r u C t u r e 9
manages millions of critical docu-ments — the kind that get sent by certified mail — each month. The company provides services from two mirrored operations centers in southern California and uses two colocation facilities for backup.
As its business grew, the compa-ny’s IT infrastructure became cumbersome. Different types of servers were needed for Web inter-faces, proprietary applications and databases, and that led to less-than-optimized use of systems and higher management and mainte-nance costs, says Bart Falzarano, chief information security officer at Walz Group.
“The challenge we were having was trying to scale out the infra-structure technology to meet our business growth requirements,” Falzarano says.
The firm’s distributed storage was difficult to scale up — as well as to manage and back up — and that affected application performance. The IT team was forced to focus on infrastructure support rather than improving products and services, Falzarano says.
Company management recog-nized that an IT revamp was need-
ed, and in 2009 opted for Cisco’s Unified Computing System (UCS) and NetApp’s FAS storage systems.
UCS combines computing, networking, storage access and VMware server virtualization into a single offering. At Walz Group, the implementation consists of five chassis housing 40 blade servers, with some configured as VMware
in a nutshell
the primary benefits of converged
infrastructures are:
Faster hardware deployment and
reconfiguration times.
Simpler management.
dynamic configuration and reconfiguration of resources, which
provides a foundation for cloud computing.
03IDGEconv_1data.indd 9 8/24/11 1:53:06 PM
1 0 C o n v e r g e d I n f r a s t r u C t u r e | d ata c e n t e r m a n a g e m e n t
ESX hosts to support customer-facing Web interfaces and propri-etary document-management applications.
To provide for disaster recovery, Walz Group installed three UCS chassis in its main colocation facil-ity and two more in other company locations. The firm’s IT depart-ment manages all blade servers from a single system, the Cisco UCS Manager, and centrally manages all NetApp storage, which houses about 80 terabytes of data.
In essence, the converged server and storage infrastructure provided the foundation for a private cloud. Walz Group also created a separate
partition on UCS (and on NetApp storage) that allows the company to offer a reve-nue-generating cloud service to its customers.
Among the biggest bene-fits of the unified computing system are greater avail-ability and the ability to move application workloads quickly. Prior to the move to converged infrastructure, Walz needed to schedule “outage windows” in order to upgrade servers and stor-age. Now, IT uses VMware
vMotion to move virtual machines to other blade servers, make any needed changes, and then move the virtual machines back without interrupting service to customers.
Using UCS Manager, Walz can configure a new server blade to support a Web portal or application in about 15 minutes, compared with four hours in the old architecture.
Before the move to a converged infrastructure, Falzarano adds, Walz was experiencing a hardware failure once or twice a month, but now such failures occur once a year or less.
The new environment has helped
“ With fewer physical servers, maintenance and licensing costs have gone down about 30%, while power and cooling costs have dropped 40%.”
C h u C k s h m ay e l , v I C e p r e s I d e n t o f I n f r a s t r u C t u r e a n d s e C u r I t y, s I rva I n C .
03IDGEconv_1data.indd 10 8/24/11 1:53:06 PM
d ata c e n t e r m a n a g e m e n t | C o n v e r g e d I n f r a s t r u C t u r e 1 1
the firm decrease capital costs by reducing the number of servers, switch ports and cables in the data center. UCS “offers the highest density of compute and memory per rack unit,” Falzarano says, “and that has resulted in a 50% rack reduction in our data centers.”
So far, Walz Group has consoli-dated about 40 servers onto six VMware ESX hosts. One result: IT labor costs are lower. Prior to moving to the new environment Walz needed seven people to manage its infrastructure full time. Now only three staffers are needed.
Overall, the move to a converged infrastructure is looking like a good move for the company. “We’ve been in production with this technology for over a year and a half and we’ve seen huge operational efficiency improvements, greater ability to scale, security improvements, [better] business continuity plan-ning and improved disaster recov-ery,” Falzarano says.
Simplifying IT managementSirva Inc., a Westmont, Ill., provider of corporate relocation services, also has moved to a converged infrastructure. In recent years, the
company went through several strategic acquisitions to expand its business, and recognized the need to simplify its IT operations and consolidate its four data centers.
“To do so, we needed to update our infrastructure to reduce the number of physical servers we had in the data centers, minimize hardware movement and reduce hardware failure risks,” says Chuck Shmayel, vice president of infra-structure and security. “We were hitting roadblocks, because our efforts to minimize IT expenses, while improving operations, proved to be a near-impossible task with the infrastructure we had at the time.”
After evaluating a number of vendor products, the company opted to consolidate its data center with Hewlett-Packard’s Converged Infrastructure offering, which consists of HP hardware and virtu-alization and management soft-ware.
The IT team at Sirva virtualized hundreds of servers using VMware vSphere running on HP ProLiant blade servers. “On the storage front, we replaced some of our older NetApp storage with HP [virtual
03IDGEconv_1data.indd 11 8/24/11 1:53:06 PM
1 2 C o n v e r g e d I n f r a s t r u C t u r e | d ata c e n t e r m a n a g e m e n t
arrays], because we found HP stor-age easier to manage and expand and control the cost,” Shmayel says.
The company has seen a number of benefits from its data center consolidation and move to a converged infrastructure — most notably, lower IT costs. “With fewer physical servers, mainte-nance and licensing costs have gone down about 30%, while power and cooling costs have dropped 40%,” Shmayel says. Moving to the HP storage arrays brought a 30% reduction in stor-age costs as well as faster access to data, he adds.
There were other benefits. Using virtual machines reduced the time needed to provision a new server to less than four hours, rather than the four to six weeks it took previ-ously. The company reduced its physical server count from 600 to 130, and achieved a tenfold reduc-tion in server racks, from 30 to three. It also cut tape costs by 50% annually, and the new system allows Sirva to automatically load-balance virtual machine resources across the blade chassis.
The challenges of moving to the new environment “were mostly
related to learning the new hard-ware and the new tools,” Shmayel says. “The way we dealt with it is by taking the time to get training. We arranged for training by HP online, on-site and hands on. The HP engi-neers played a key role in knowl-edge transfer as well.”
Boosting efficiency, lowering costsAnother company, Slumberland Inc., a privately held furniture retail-er based in Little Canada, Minn., in 2009 began looking to replace aging hardware in its data center, which supports end users in about 120 stores located in 11 states.
The company’s IT department investigated new data center archi-tectures that would support busi-ness growth. “We were looking for a highly standard, highly secure and highly scalable solution that was very cost-effective,” says Jamie Page, CIO at Slumberland.
Page says the company decided to implement Cisco’s UCS because it provides the technologies and capabilities — such as virtualiza-tion and low-cost scalability — that the retailer needs to support the expected growth in data. Slum-
03IDGEconv_1data.indd 12 8/24/11 1:53:06 PM
d ata c e n t e r m a n a g e m e n t | C o n v e r g e d I n f r a s t r u C t u r e 1 3
berland’s UCS envi-ronment includes four chassis, each with multiple server blades. A test of the retailer’s environment showed that the ideal density was 20 virtu-al machines per blade, Page says.
The company chose Microsoft’s Hyper-V virtualization soft-ware because of its small footprint and the IT staff’s familiarity with Microsoft software. The UCS hosts a variety of operating systems used at Slumberland, including Microsoft Windows Server 2003 and 2008 and Red Hat Enterprise Linux. It also hosts Microsoft SQL Server, used for data warehouses and a homegrown ordering appli-cation; and Cisco Unified Communi-cations Manager, used for IP tele-phony.
To manage the environment, the company uses Cisco UCS Manager software along with Microsoft Windows Deployment Services.
The new environment is more efficient than the previous one, Page says. “We are able to put far
more users on each half-width blade and whole-width blade than before,” he says. “With virtual-ized servers and [a storage-area network], it’s a very low cost of ownership.” The company is saving about $1,700 on each logical server it deploys.
Overall, the company has a simpler IT infrastructure that’s far more scalable. “We still have separate servers and storage-area network, but connectivity [and integration] between them has become simpler” because of the converged infrastructure, Page says. n
B o B V i o L i n o is a freelance writer in mass-
apequa park, n.y. you can contact him at
We still have separate servers and storage-area network, but connectivity between them has become simpler.j a m I e pa g e , C I o s lu m B e r l a n d
03IDGEconv_1data.indd 13 8/24/11 1:53:06 PM
14 C o n v e r g e d I n f r a s t r u C t u r e | N e t w o r k s
networks key to
puzzleconvergence
03IDGEconv_2net.indd 14 8/24/11 1:54:49 PM
N e t w o r k s | C o n v e r g e d I n f r a s t r u C t u r e 1 5
upgrading the
network helps
get the most
out of server
infrastructure,
simplifies wiring,
and paves the
way for cloud
By Maria korolov
he move to converged
computing and storage is
well on its way, but the leg-
acy networks at many companies
are an obstacle to modernization.
Network convergence helps compa-
nies support the demands of server
virtualization, and can become the
first step on the road to the adoption
of network fabrics and private cloud
architectures.
Tnetworks key to
03IDGEconv_2net.indd 15 8/24/11 1:54:52 PM
1 6 C o n v e r g e d I n f r a s t r u C t u r e | N e t w o r k s
Full-scale convergence requires ripping out existing networks and replacing them with new, mostly untested and unfinished technol-ogy. Instead, most companies are
taking a more gradual approach to convergence — one section of their network at a time, or one compo-nent or one layer at a time.
But there are some that want to get it done in a hurry. Stephens Inc., a privately held investment bank in Little Rock, Ark., consolidated its data center network in one big step. “It was a forklift, over a week-end,” says Vice President Matthew McEwen. The company pulled out its old hardware — 26 top-of-rack switches and associated cabling and 1 Gigabit backbone — and replaced it with two Enterasys S-Series 10-Gigabit switches.
Enterasys’ Data Center Manager enables Stephens to centrally manage its network of 3,000 devices. Stephens is also using Enterasys’ NMS Policy Manager
to improve application delivery to more than 1,000 users throughout the U.S.
Stephens is already seeing signif-icant savings, McEwen says. “The
project will pay for itself by the end of the year — less, when you take into account manpower costs. Now I don’t have to have more than one administrator during the configura-tion state,” because the network can automatically roll out commu-nication and security policies. “As servers move around, I no longer require a network security level administrator to verify the servers are talking to the servers they need to be talking to. I’m able to leverage a much smaller staff than most, and the staff that I do have can work on upcoming projects and tuning the network, rather than troubleshoot-ing problems on the network.”
Stephens still has a traditional three-layer network, McEwen says. And there’s still Fibre Channel on the storage side, which is scheduled
“ The project will pay for itself by the end of the year — less when you take into account manpower costs.”
M at t h e w M C e w e n , v I C e p r e s I d e n t, s t e p h e n s I n C .
03IDGEconv_2net.indd 16 8/24/11 1:54:52 PM
N e t w o r k s | C o n v e r g e d I n f r a s t r u C t u r e 1 7
to be upgraded to iSCSI in the next two to three months.
Converging, simplifyingNetwork convergence can help a company reduce the complexity of its network infrastructure, and reduce both the number of network layers and the number of data pipes.
With a typical server, there are cables going to the internal produc-tion environment, a management network, a backup network, a stor-age network, the VMware manage-ment network, the VMware VMotion network and a fault toler-ance network, says Jon Toor, vice president of marketing at network-ing technology vendor Xsigo Systems. “You can get to 20 cables per server in a big hurry.”
Because each cable is physically isolated from the others, a sudden increase in traffic can cause prob-lems. “You’ve got no way of sharing that load among the other pipes, the other pipes are sitting around idle, and that one server is now a bottleneck,” Toor says.
“If you’re going to truly get to the cloud, you need to virtualize everything,” Toor says. And that includes switching out the 10 or 20
separate cables and replacing them with one big pipe — plus a second for backup — and having multiple virtu-al networks, logically isolated from one another, all running through that one pipe, each taking up more or less bandwidth as needed.
That’s what South Africa’s Investec Bank Limited achieved when it installed Xsigo switches and interface cards, reducing the total number of cables from 1,980 to 260.
“ If you’re going to truly get to the cloud, you need to virtualize everything.”
j o n to o r , v I C e p r e s I d e n t o f M a r k e t I n g , x s I g o s y s t e M s
03IDGEconv_2net.indd 17 8/24/11 1:54:55 PM
1 8 C o n v e r g e d I n f r a s t r u C t u r e | N e t w o r k s
“We didn’t go all the way to a network fabric,” says Suren Naidoo, manager of the bank’s shared services IT and global infrastruc-ture integration team. “Before this, we had a three-tier network – a full Cisco network. Cisco chassis, Cisco blades, Cisco switches. Our core is still Cisco, but between that and the servers is the Xsigo layer.”
As a result, the three-tier network has been reduced to two layers. “This is a simplified approach,” he says.
Each Dell server now has two Xsigo Director cards, instead of the previous structure, where each
server required 13 network inter-faces and two host-bus adapter cards. The total number of network interfaces and ports dropped from 572 to 52, Naidoo says.
Now a new server takes just two hours to deploy, compared to the three weeks it took previously, help-ing the company get to market fast-er with new applications, he adds.
Longer term, the bank plans to modernize the networks in its loca-tions in Australia and the U.K., and find a way to use one of those data centers for disaster recovery. “If we shut down the disaster recovery center, that’s a cost savings for us of $100 million in total cost of owner-ship,” he says.
It’s all part of the bank’s move toward a private cloud, he says. “That’s 100% in line with the discussions we have right now. The private cloud is the way we’re tran-sitioning.”
Moving toward mesh: One component at a time Concur Technologies is a provider of corporate travel and expense management applications based in Redmond, Wash. The company originally began looking at server
“ Our architecture is a lot flatter than it used to be, easier to maintain and a lot more resilient than in the past.”
C h r I s j e n n I n g s , d I r e C to r o f e n g I n e e r I n g s e rv I C e s , C o n C u r t e C h n o lo g I e s
03IDGEconv_2net.indd 18 8/24/11 1:54:56 PM
N e t w o r k s | C o n v e r g e d I n f r a s t r u C t u r e 1 9
virtualization as a way to save space and capital expenses, and network convergence is part of that puzzle.
“We’re taking a cautious approach,” says Chris Jennings, director of engineering services. “We’re replacing components one at a time.” The goal is to have a network mesh fabric. “But at this point, we don’t feel it’s baked enough.”
Instead, the company is replacing its old Spanning Tree top-of-rack switches with Brocade FCX series switches and, more recently, the Brocade VDX. HP Virtual Connect Flex-10 is used to connect the servers to the Brocade switches at
10Gbps. The Brocade top-of-rack switches, in turn, connect to a Brocade NetIron MLX core switch. In addition, the company uses a network management appliance from ExtraHop.
“Our architecture is a lot flatter than it used to be, easier to main-tain and a lot more resilient than in the past,” Jennings says.
The company is now about 80% of the way through the upgrade process, he says, and the physical footprint has been reduced by 40% to 50%. In addition, the number of cables has been substantially reduced.
According to HP’s Duncan Camp-bell, vice president for worldwide marketing in converged infrastruc-ture, a “rip and replace” approach to network convergence is just not realistic.
“Our view is that we have to work with what’s installed, provide
both a building block approach and a transformational approach.” As a result, customers can gradu-ally transition from a three-tier hierarchical architecture to a flat converged fabric.
If the migration to fabrics was abaseball game, then the pitcher
hasn’t even finished warming up. “The game hasn’t even started yet.” Z e u s k e r r ava l a , ya n k e e g r o u p
03IDGEconv_2net.indd 19 8/24/11 1:54:58 PM
2 0 C o n v e r g e d I n f r a s t r u C t u r e | N e t w o r k s
“Each step you take, you’re going to get more advantage on price and on performance,” Jennings says.
The piece-by-piece approach is the best way to go right now, agrees Yankee Group’s Zeus Kerravala. “It is a prudent way to go because the technology is so new. For where we are in the market today, I think that’s a good approach. It offers a safer migration path.”
The only vendor to have a ship-ping fabric is Brocade, Kerravala says. “Cisco has a large pipeline with a number of different fabric tech-nologies. And Juniper has done a good job with raising the visibility of what QFabric is, but they need to get the product out.”
The industry needs to see some working deployments of network fabrics, Kerravala says, and proof they improve traffic flows and remove bottlenecks.
It is still in the early days, he says.
Vendors are positioning themselves more for mind share than anything else. If the migration to fabrics was a baseball game, then the pitcher hasn’t even finished warming up, he says. “The game hasn’t even started yet.”
Kerravala estimates the adop-tion rate of modern, fabric-style networks at less than 3%.
At some point, however, the technology will be ready, and companies should get more aggressive about moving to a full-scale converged, single-layer, mesh fabric network, he says.
The network’s virtual future: CloudNetwork fabrics will ultimately help simplify adoption of cloud computing.
Traditional networks are relatively inflexible, making it
“ Developers and end users have gotten frustrated with IT and have gone externally to spin up servers. That has led to risk to organizations, with data sitting on unqualified assets.”
s I M o n j o h n s o n , d I r e C to r o f d ata C e n t e r s e rv I C e s , g l a s s h o u s e
03IDGEconv_2net.indd 20 8/24/11 1:55:00 PM
N e t w o r k s | C o n v e r g e d I n f r a s t r u C t u r e 2 1
expensive and time-consuming to deploy servers, says Simon John-son, director of data center services at GlassHouse, an IT consulting firm in Framingham, Mass.
The result: adoption of public cloud resources. “Developers and end users have gotten frustrated with IT and have gone externally to spin up servers,” Johnson says. “That has led to a risk to organiza-tions, with data sitting on unquali-fied assets.”
Converged networks powered by network fabrics will enable IT to meet the dynamic needs using internal resources.
“Network convergence means I can, as an end user, get support for my application at a cheaper price point,” Johnson says. “I don’t have to wait so long. I get a better quality of service. I get collapsed protocols and a simpler operating environment from an IT perspective, and can get to market more quickly.”
At the same time, converged networks make it more feasible to view public cloud resources as an extension of a private cloud when it makes sense from a business and security perspective.
The convergence of networks and the move to fabrics is the logical
next step for companies look-ing to take their server and storage virtualization efforts to the next level as they contemplate a future built around the cloud. How buyers get there will depend on their legacy infrastructure and current business demands, but there seems to be little question about the direction forward. n
M a r i a ko r o lov is a freelance busi-
ness and technology writer in Massa-
chusetts. she can be reached at maria
@ trombly international.com.
03IDGEconv_2net.indd 21 8/24/11 1:55:01 PM
2 2 C o n v e r g e d I n f r a s t r u C t u r e | H o w T o G u i d e
a hands-on guide to
a typical midsize data center gets
a makeover — providing a detailed
example of how to deploy a
modern, converged infrastructure
By MaTT PriGGe
new wind is blowing in the data center — and it’s not the water-cooled air we’re used to.
The rapid adoption of all things converged is changing the basic character of the data cen-
ter network. The wind is pushing low-bandwidth, purpose-specific application and storage networks out — and leaving multi-purpose, high-bandwidth networks in their place.
Nearly limitless scalability and agility is the biggest benefit of infrastructure convergence. By combining all I/O into a single network, you have a single pool of network resources to draw from, yielding vastly better utilization and efficiency.
convergence
A
03IDGEconv_3benefits.indd 22 8/24/11 2:41:50 PM
H o w T o G u i d e | C o n v e r g e d I n f r a s t r u C t u r e 2 3
convergence
So how do you transition from your existing data center to converged infrastructure? The best way to explain the process is to show the “before and after” transformation step by step. We’ll start with a typical data center in a midsize company.
The lay of the landImagine that you’re the infra-structure manager for a midsize company. In your data center you have around 300 servers. Of these, approximately 200 are virtualized and the remainder are still physical.
The virtualized servers are split across two virtualization clus-ters — the newer of which is based on relatively recent blade servers while the older ones are fairly typi-cal rack-mount servers. The physi-cal servers are an odd assortment of older machines that haven’t been virtualized yet and some newer boxes for which virtualization hasn’t been an option in the past (either due to load or lack of support from software vendors).
The data network is comprised of redundant top-of-rack Ether-net switches that are backhauled
by multiple 1Gbps links to a pair of data center distribution switches. Most physical servers attach to the network via a pair of 1Gbps Ether-net links while the virtualization hosts have four or more each.
On the storage side, two mid-range Fibre Channel SANs attach to a traditional 4Gbps Fibre Channel SAN fabric. One of these SANs is nearing end of life while the other will remain in service for a few more years. The Fibre Channel network is made up of four switches — two in each of two redundant fabrics.
This typical midsize data center configuration is ripe for conver-gence — and can be transformed without tearing everything out in one shot. The phases laid out here could be implemented in short order, budget dollars and human resources permitting — or they might be completed over several years in manageable chunks.
Either way, it’s critically impor-tant to know what you want the result to look like before you start. That’s because the early steps you decide to take will determine the amount of flexibility you have in dealing with changes down the road.
03IDGEconv_3benefits.indd 23 8/24/11 2:41:51 PM
2 4 C o n v e r g e d I n f r a s t r u C t u r e | H o w T o G u i d e
Virtualizing to the hiltServer virtualization is key to convergence design. Virtualization, after all, will be the main consumer of the massive performance poten-tial inherent in a converged switch-ing infrastructure.
Just as server virtualization allows you to run many virtual servers on a single high-perfor-mance physical host — and do away with a ton of power-sucking server hardware — it also eliminates a ton of lower-bandwidth Ethernet and Fibre Channel ports. By consoli-dating those loads on virtualiza-tion hosts, you can give your new 10Gbps network links a much better run for their money, while decreas-ing the number required. The fewer physical servers left at the end of this process, the better — first because it will cost less, but also because it will be easier to manage.
So, let’s say that out of the 100 physical servers left in the data center, 50 of them are easy virtu-alization candidates, 40 couldn’t be virtualized when they were installed due to load concerns and 10 run apps that aren’t supported in
a virtualized environment. This would be a great time to
re-evaluate the 40 that were configured as physical servers to determine if those load constraints really exist — especially in the face of newer, faster server hardware and improved storage bandwidth/latency. At this stage you should also re-evaluate the support constraints on the remaining 10 servers — many staunchly anti-virtualization software vendors have suddenly changed their tune even over the past year or so (amaz-ingly, this is true even in the tightly controlled healthcare sector).
Regardless of how many of the remaining physical servers we’re able to virtualize, it’s unlikely you’ll have enough available capacity in the existing virtualization clus-ters to absorb it. So this probably means you will need to make some substantial additions to the virtu-alization infrastructure to improve the penetration of virtualization.
This presents a challenge that crops up often during sea changes in IT: It’s difficult to turn the corner toward a new technology without biting off significant capital costs. You can either make the decision
03IDGEconv_3benefits.indd 24 8/24/11 2:41:51 PM
H o w T o G u i d e | C o n v e r g e d I n f r a s t r u C t u r e 2 5
to do more of what you’ve already been doing — deploying blade-based virtualization hosts with separate, non-converged networking and storage hardware — or move immedi-ately to a fully converged network.
The first case demands a smaller immediate capital expense, of course. But down the line it will prove to be less efficient than if you had reconfigured the hosts to run on a converged network. The second case will require a much larger capital investment, but it will be more cost-efficient in the long run.
Converging the networkThe first thing to get in place on the path to convergence is a pair of DCB (data center bridging) switches. These switches will initially be used to replace your data center distribu-tion switches.
Almost any DCB switch available today ships with a varying number of SFP+ ports, which can gener-ally accommodate 10Gbps Ethernet SFP+ modules, 4Gbps and 8Gbps Fibre Channel modules and 1Gbps Ethernet SFP modules. Some switch vendors dedicate a specific set of
ports for Ethernet or Fibre Channel, while others allow you to mix and match at will (sometimes subject to additional licenses). When choosing a platform with a fixed port config-uration, make sure you avoid paint-ing yourself into a corner — where you’re just shy one native Fibre Channel port, for example, but have 10 unused Ethernet ports.
An appropriately spec’d DCB switch gives you the flexibility to reabsorb all of the 1Gbps Ethernet uplinks from top-of-rack switches, consolidating the core switching for the entire data center onto a single platform. These data center core switches can be attached to the existing Fibre Channel fabric via dedicated 4Gbps or 8Gbps Fibre Channel E-Port (ISL) links — essen-tially extending the two redundant fabrics into the two new converged switches as if they were a third member of each SAN fabric.
After some thorough burn-in and testing, the next order of busi-ness is to start drawing the largest networking and storage consum-ers onto the converged network. One of the largest consumers — if not the largest — will be the blade-based virtualization environment.
03IDGEconv_3benefits.indd 25 8/24/11 2:41:51 PM
2 6 C o n v e r g e d I n f r a s t r u C t u r e | H o w T o G u i d e
Depending upon what kind of blade chassis you’re dealing with, you may find that the blades you already own and run have 10G Ethernet CNAs (converged networking adapters) built onto the motherboard. In that case, reconfiguring the blades and chassis may simply require imple-mentation of new 10G-capable inter-connect modules.
If the blades are not already equipped with 10G CNAs, then you’ll probably need both new intercon-nect modules for the chassis as well as CNA modules for the blades themselves. If you find yourself in this boat, it may end up being more cost-effective to start building a new blade environment that is config-ured for convergence from the get-go and upgrade the existing chas-sis once the blades within it have reached the end of their service life.
Using a second (or spare) blade chassis is also a good way to manage this migration without introduc-ing downtime. Blades can easily be removed from service, moved and reconfigured one at a time. This addi-tional blade chassis capacity may also prove useful as physical rack-mounted servers are replaced with blade-based servers down the line.
In either case, most 10G/FCoE blade chassis interconnects avail-able today will allow you to scale the number of 10G links that attach the chassis (and all of the blades within it) to the converged switch fabric at will. It’s generally prudent to plan on at least four to start with — providing 20Gbps of converged storage/network bandwidth to each of the redundant core switch-es for a total of 40Gbps. Chances are that this will easily outstrip the aggregate bandwidth of what was being replaced. More links easily can be added if they are required.
Once a large portion of storage consumers on the network have been either virtualized or migrated to FCoE, it’s time to collapse the existing storage network. Given that both of the existing SANs in use are Fibre Channel-only, this is simply an issue of walking them over interface by interface to native Fibre Channel interfaces on the converged data center core switches.
Once that’s completed, the only systems still attached to the tradi-tional Fibre Channel switches are the remaining physical machines that couldn’t be virtualized for whatever reason. If these servers
03IDGEconv_3benefits.indd 26 8/24/11 2:41:51 PM
H o w T o G u i d e | C o n v e r g e d I n f r a s t r u C t u r e 2 7
are nearing the end of their service life, it probably makes sense to leave them as-is until they can be replaced — preferably with blades that live side-by-side with the virtualization blades and share a similar CNA-equipped configura-tion. Alternatively, you can either install CNAs in them and attach them to the data center core inde-pendently, move their 1G and 4G Fibre Channel ports to dedicated ports on the core without replac-ing any hardware, or leave them attached to the traditional Fibre Channel switches without chang-ing anything.
In most cases, however, you’re better off removing the physical Fibre Channel switches as soon as you can — if for no other reason than to make performance issues easier to troubleshoot by providing a single point of management at the core.
Prepping for the futureAs time goes on and the data center grows, the foundations laid in migrating to a converged infra-structure truly can be realized.
Additional servers can be installed in the form of virtual
machines or additional blades without any substantive changes to the network whatsoever. As blade chassis fill, new chassis can be added and attached to the converged core, providing addi-tional capacity for blades. As the amount of data and storage band-width consumed increases, 10Gbps links can be added to the chassis. And, as more storage resources are required (replacing that aging SAN for example), they can be deployed using the same 10G-based FCoE.
A scenario like the one described here makes the benefits of conver-gence obvious. Just in this fairly limited example, 14 pieces of data and storage networking equip-ment and more than 300 cables were replaced by a single pair of switches serving just 40 connec-tions between them — all while decreasing the amount of stranded network capacity through heavy load consolidation. n
m aT T P r i G G e is an InfoWorld contribut-
ing editor and 10-year veteran of enterprise
internetworking and infrastructure. If there’s
a technology in use in the enterprise today,
chances are he’s designed it, built it, fixed it
and supported it. reach him at matt_
03IDGEconv_3benefits.indd 27 8/24/11 3:10:42 PM
2 8 C o n v e r g e d I n f r a s t r u C t u r e | S ta f f i n g
how to staff for converged IT
as convergence shakes up the data center, finding the right people who are highly versatile and cross-disciplined is key for CIos who want to stay ahead of the curve by Pam baker
03IDGEconv_4staffing.indd 28 8/23/11 6:13:18 PM
S ta f f i n g | C o n v e r g e d I n f r a s t r u C t u r e 2 9
onverged InfrasTrucTure Is becoming a hot topic of conversation as of late, albeit a relatively new one for those at the top of the IT food chain, namely CIOs. Because the concept of converged in-frastructure is so new, relatively few CIOs
have wandered in deep enough to report back on what worked and what did not. There are, however, emerging management and staffing patterns for successfully han-dling the transformation process.
As data centers have expanded and proliferated over the past few years due to increases in data and physi-cal servers, along with a rise in energy consumption and increased support costs, many companies are addressing these issues by moving toward a converged infrastruc-ture by pooling IT resources and increasing automation. Predictions abound that soon the data center as we know it will simply disappear or morph so much beyond recogni-tion that it will be something else entirely.
C
03IDGEconv_4staffing.indd 29 8/23/11 6:13:20 PM
3 0 C o n v e r g e d I n f r a s t r u C t u r e | S ta f f i n g
“The data center is increasingly viewed by those on the bleeding edge as a giant computer instead of as a building to house IT,” explains Katie Broderick, senior research analyst at IDC.
‘You are here’While a few examples of success-fully converged IT exist, such as at DreamWorks Animation, most CIOs have barely begun to think about infrastructure convergence, much less actually deploying it. More than a few are still figuring out what it means.
“Infrastructure convergence is still a loosely defined term,” says George Weiss, vice president and distinguished analyst at Gartner. Even so, he says, its essence is rela-tively clear: “The core function of the data center shifts from build and deliver capacity to delivery of services — from which stems the concept of cloud as a service infra-structure.”
It is common for IT convergence to begin in any given organization with a measured and finely paced aggregation of the more basic services, typically starting with software as a service (SaaS) and slowly moving forward.
“Since convergence applies to the server, network and storage systems as an integrated logical view, the traditional roles of server, network and storage administra-tors must also converge as software defined management solutions are layered on the hardware infrastruc-ture to exploit this integration,” Weiss explains.
As a result, staffing requirements shift away from traditional opera-tional roles and toward assembling groups of people with the knowl-edge and skills to understand and handle increasing levels of conver-gence. While such a shift in skill sets is necessary, it typically isn’t popular.
“Many cultural and political barri-
“ The data center is increasingly viewed by those on the bleeding edge as a giant computer instead of as a building to house IT.”
K at I e b r o d e r I C K , s e n I o r r e s e a r C h a n a ly s t, I d C
03IDGEconv_4staffing.indd 30 8/23/11 6:13:21 PM
S ta f f i n g | C o n v e r g e d I n f r a s t r u C t u r e 3 1
ers exist in launching wholesale job role change in IT as IT specialists attempt to hang on to their exper-tise for job security,” Weiss warns.
For the moment, though, IT staff-ing trends overall remain static with just a few new positions rising in the ranks.
Talent checklist“Positions in application manage-ment and performance, security, configuration management, archi-tectural design and database admin
have not seemed to have been radically altered, but their
need to expand their awareness and expertise to higher levels of mobil-
ity and dynamic real time infrastructure are putting
pressures to upgrade and rede-fine these positions,” Weiss says.
The skills that are most in demand are those related to how infrastructure convergence is done in the first place: through virtual-ization.
“Skills in virtualization continue to be red-hot,” says Bill Brydges, managing director of technology services at MorganFranklin, a finan-cial and IT consultancy. “In addition
to ongoing needs for hardware and OS optimization skill sets, there is an increased focus on storage virtu-alization that is creating new roles for IT staff.”
“The ability to manage vendors and external provided services is also critical, as these services are increasing across the enterprise,” Brydges adds.
However, infrastructure conver-gence is still emerging in prac-tice, so not all staffing needs are readily evident.
“Standard industry job titles and descriptions for the consolidated func-tions are yet to emerge,” says John Rivard, research director at Gartner.
Still, movement toward conver-gence is steadily growing momen-tum, fueled as it is by the compel-ling need to control costs and manage complexity.
“ Skills in virtualization continue to be red-hot.”
b I l l b ry d g e s , m a n a g I n g d I r e C to r o f t e C h n o lo g y s e rv I C e s , m o r g a n f r a n K l I n
03IDGEconv_4staffing.indd 31 8/23/11 6:13:22 PM
3 2 C o n v e r g e d I n f r a s t r u C t u r e | S ta f f i n g
“Converged infrastructure, along with consolidation, ‘lights out’ data centers and cloud computing gener-ally reduce the number of positions required — but the ones that remain require higher-level technology planning and strategy manage-ment skills,” says Faisal Hoque, founder and CEO of BTM Corpora-tion and author of “The Power of Convergence.”
“Thus we will see an increased need for ‘business architects’ as opposed to the pure technologists who ran data centers in the past,” he adds.
New data center management schemesData center management organi-zation also is being reconfigured in order to handle the increased complexity in IT, and not just for deploying infrastructure conver-gence. Several new data manage-ment models are emerging.
“They are being organized, in some bleeding edge cases, around workloads rather than around a server team, a storage team, a networking team and a facilities team,” IDC’s Broderick says. “In other cases they are being orga-
nized around a business unit.”Bottom line, data center manage-
ment models are evolving to meet a focus change from technol-ogy management to information management.
“Converged infrastructure is reinforcing the trends toward busi-ness service focus, adaptation of agile practices in I/O and greater end-to-end process management,” Gartner’s Rivard says. “Knowing the needs of your business and the characteristics of the workloads supporting their critical business processes combined with increased infrastructure agility is key.”
Thus, developing an effective management model requires an accurate anticipation of what the desired goal looks like and the steps IT will need to take to reach that goal.
“ Converged infrastructure is reinforcing the trends toward business service focus.”
K at I e b r o d e r I C K
03IDGEconv_4staffing.indd 32 8/23/11 6:13:23 PM
S ta f f i n g | C o n v e r g e d I n f r a s t r u C t u r e 3 3
“The key new requirement is to plan out an enterprise business architecture — and a supporting technical architecture — in advance, which you can use to map the converged infrastructure that you are moving to,” BTM’s Hoque says. “Without this there is no assurance that the data center will support the evolving needs of the business.”
In any case, the best manage-ment models will be highly custom-ized to the organization’s actual needs, which will typically incorporate hybrid envi-ronments and hybrid skills in the IT staff.
“The reality is that most IT organiza-tions will not move wholesale to a public cloud, private cloud, internal converged, all outsourced or all hosted model,” Broderick says. “There will be a mix based on workload and application appropriateness for each environ-ment.”
Points to ponderThere is more to consider in mapping out your staffing needs than how best to service business users and manage a consolidated infrastructure.
“For a CIO to be considered leading-edge, they must keep any differentiating proprietary informa-tion/processes in-house,” says Jim Venglarik, CIO of China’s leading IT and software outsourcing provider, Bleum. “This means that there will
be a continued need for staffing to accommodate development of
these processes.”Such challenges will
continue to arise and the CIO must remain vigilant to guard
against overlooking key impact points. That, and the diminishing
need for technology trouble-shoot-ing, will in turn affect how the role of CIO is defined.
“With an increase in streamlin-ing processes through infrastruc-ture convergence, the future role
“ The reality is that most IT organizations will not move wholesale to a public cloud, private cloud, internal converged, all outsourced, or all hosted model.”
K at I e b r o d e r I C K
03IDGEconv_4staffing.indd 33 8/23/11 6:13:24 PM
3 4 C o n v e r g e d I n f r a s t r u C t u r e | S ta f f i n g
of the CIO may very well change,” Venglarik says.
“Although there will always be a need for senior management to make tough decisions, including budgetary and contractual, can the CIO’s role become part time?” he asks. “Can it become virtual? Does the CIO of the future need to be anything more than just a guiding hand to make sure things operate according to plan?”
Use case: DreamWorks AnimationIt’s ironic that CIOs spent decades keeping the lights on only to find they need to turn the lights out. “Lights out” means a locked down,
fully automated data center that requires no human intervention at all; “lights out” data centers are today the holy grail of infrastruc-ture convergence.
“Our goal is to operate our off-site data centers as lights-out facilities,” says Ed Leonard, CTO of Dream-Works Animation. “Of course this requires a significant amount of systems engineering work to make this happen.”
DreamWorks Animation, Leon-ard says, finds the idea of infra-structure convergence to be an important step in packaging its infrastructure needs into module-size chunks. “Once we have these chunks, we can begin the conver-sation about where and when to house these chunks which quickly turns into a conversation about flexibility to scale quickly.”
The implications on staffing, he says, are twofold. First, he was able to leverage the skill sets of data center staff across all of Dream-Works’ locations without having to
“ Our goal is to operate our off-site data centers as lights-out facilities.”
e d l e o n a r d, C to, d r e a m w o r K s a n I m at I o n
03IDGEconv_4staffing.indd 34 8/23/11 6:13:26 PM
S ta f f i n g | C o n v e r g e d I n f r a s t r u C t u r e 3 5
“scale up operations staff to keep pace with a fast-growing demand for compute and storage.” In other words, he found the skills he need-ed in existing staff and did not need to hire any extra bodies in order to deploy infrastructure convergence. Second, he says, his IT staff is now free to focus on the business end rather than the technical end of operations.
“For DreamWorks, it’s less about new positions being created than a focus on ‘cloud-enabling’ skills within the teams,” Leonard says. “We’re at the forefront within our industry in leveraging cloud for HPC workloads and have been doing that successfully and in increas-ing amounts over the last several years.”
Leonard says the top skills needed today are virtualization, SOA-based concepts such as a service/enterprise bus, and service-level management. “These aren’t really new, but perhaps now more mature and better organized,” he says. “A solid systems infrastruc-ture foundation needs to be there, but now our time is spent higher in the stack, with high-value targets.”
Actions to take nowThe promise in converged infra-structure is more automation, greater agility, fewer costs and a much smaller IT staff. Successfully achieving this end relies heavily on finding the right talent, that is to say, personnel who are highly adaptable and cross-disciplined. “Infrastructure convergence is driv-ing CIOs to hire and develop ‘versa-tilists,’ i.e. individuals with broad and deep knowledge across a wider set of infrastructure components,” explains Gartner’s Rivard.
To prepare for the changes ahead,
“ It’s less about new positions being created than a focus on ‘cloud-enabling’ skills within the teams.”
e d l e o n a r d
03IDGEconv_4staffing.indd 35 8/23/11 6:13:26 PM
3 6
regardless of the model or hybrid thereof that you deploy, Rivard recommends CIOs take the follow-ing actions immediately to develop IT staff:
1. Drive increased collaboration across the areas of platform, stor-age and network.
2. Identify and develop the “versatilists” in your organizations who will have a broad and deep understanding across the infra-structure components.
3. Establish or expand the busi-ness services culture and agility in your team. n
Pa m b a k e r has written hundreds of
articles in leading technology, business and
finance publications. she has also authored
eight books and an award-winning docu-
mentary on paper-making. she is a member
of the national Press Club (nPC), society of
Professional Journalists (sPJ) and the Inter-
net Press guild (IPg). follow her on twitter:
@bakercom1.
i D g e n t e r P r i S e492 old Connecticut Path P.o. box 9208 framingham, ma 01701-9208 (508) 879-0700 fax (508) 875-4394
President/CeOmichael friedenberg (508) 935-4310
executive assistant to the President/CeOPamela Carlson (508) 935-4274
S a l e Sbob melk senior vice President/group Publisher(415) 975-2685
brian glynn senior vice President, digital sales(508) 935-4586
sean weglage senior vice President, digital sales(415) 978-3314
C i r C u l at i O ndebbie winders vice President(508) 820-8193
C u S t O m S O l u t i O n S g r O u PCharles leesenior vice President(508) 935-4796
D i g i ta l S O l u t i O n S g r O u Pgregg Pinsky senior vice President/general manager (508) 271-8013
e D i t O r i a lJohn gallant senior vice President/Chief Content officer(508) 766-5426
e v e n t Sellen daly senior vice President (508) 935-4273
f i n a n C e & O P e r at i O n Smatthew C. smith senior vice President/Coo (508) 935-4038
H u m a n r e S O u r C e SPatty Chisholmsenior vice President(508) 935-4734
m a r k e t i n gsue yanovitch vice President (508) 935-4448
03IDGEconv_4staffing.indd 36 8/23/11 6:13:29 PM
Optimizing Applications in a Virtualized Infrastructure
Moving to a virtualized and more agile data center infrastructure can reduce CapEx and OpEx, and streamline application deployment and delivery. Moving applications from dedicated hardware resources to virtualized software, however, can also increase resource constraints and negatively affect application performance and availability on the network. F5 offers multiple products and solutions that address application optimization and availability for virtualized infrastructure.
Virtualized Applications and the Network
Analyst firm Nemertes Research reports that “49.4 percent of enterprise applications and 45.6 percent of mission-critical applications now run on virtual servers.” There’s no question that enterprises are embracing the cost savings and agility of virtual machines and servers; however, as more and more applications are moved to virtual platforms, other considerations become more critical for IT. Provisioning new machines, managing resources, and ultimately creating an agile orchestration and provisioning become projects that require integration with the entire IT infrastructure. This can lead to heightened resource management and application performance challenges.
One key benefit of a virtual infrastructure is its flexibility. Once core parts of the data center have been virtualized, the infrastructure’s capabilities begin to expand beyond what’s available in a physical infrastructure. The ability to scale infrastructure in real time based on resource demand is not easily achievable with physical infrastructure. But the ability to scale is also one of the challenges of virtualized infrastructure: as application services begin to rely on abstracted resources, computing and networking resources can rapidly become depleted and negatively affect application performance.
Highly tuned application optimization tools—both in computing and networking resources—typically only focus on the application data and ignore the rest of the delivery system: the virtual platform, the virtual and physical networks, and the behavior differences between applications running in a physical environment and those running in a virtual environment.
F5’s Application Delivery Network (ADN) and optimization solutions integrate directly with virtual platforms and focus on optimizing applications and the network within the virtual infrastructure. This level of integration allows deep application awareness and optimization to be applied at the virtual platform level, drastically increasing application performance.
Solution Profile | Virtualization
Key features
•Multi-Hypervisor Support—Integrates with multiple virtual platform solutions
•Service Offloading—Allows more efficient application resource utilization
•Application Availability—Monitors application availability using application-aware health checks
•Virtualized ADN—Provides holistic Application Delivery Networking solutions for virtualized infrastructure
Key benefits
•Flexible Deployment Models—Tunes application optimization and availability solutions for virtualized environments
•Support for Virtual and Physical Infrastructure—Optimizes Application Delivery Networking infrastructure for physical and virtual machines
•Efficient Resource Management—Utilizes virtualized resources for applications more efficiently
• Integrated Platform Management—Manages virtual and physical networks together as one cohesive Application Delivery Network
Securing the CloudCloud computing has become another key resource for IT deployments, but there is still fear of securing applications and data in the cloud. With F5 devices, you can keep your most precious assets safe, no matter where they live.
by Peter Silva
Technical Marketing Manager
F5 White Paper
Cloud Balancing: The Evolution of Global Server Load Balancing Cloud balancing evolves global server load balancing from traditional routing options based on static data to context-aware distribution across cloud-based services.
by Lori MacVittie
Technical Marketing Manager, Application Services
F5 White Paper
The Optimized and Accelerated Cloud As more organizations begin moving applications into the cloud, congestion will become an increasingly critical issue. F5 offers solutions for optimizing and accelerating applications in the cloud, making them fast and available wherever they reside.
by Alan Murphy
Technical Marketing Manager
F5 White Paper
Optimizing Applications in a Virtualized Infrastructurepublished by F5
Virtualized infrastructures are highly dynamic and increasingly automated. By comparison, the network delivering applications to users tends still to be static, fragile and manually configured. This mismatch can lead to erosion of the benefits of virtualization, errors, increased complexity, poor performance and even downtime. The network needs application delivery optimization that can participate in the automated orchestration of virtual infrastructure to improve and guarantee performance and availability.
http://resources.
idgenterprise.com/original/
AST-0045214_optimizing_
applications_virtualized_
infrastructure.pdf
Cloud Balancing: The Evolution of Global Server Load Balancingpublished by F5
Cloud balancing evolves global server load balancing from traditional routing options based on static data to context-aware distribution across cloud-based services.
http://resources.
idgenterprise.com/original/
AST-0001237_F5_Cloud_
Balancing.pdf
The Optimized and Accelerated Cloudpublished by F5
As more organizations begin moving applications into the cloud, congestion will become an increasingly critical issue. F5 offers solutions for optimizing and accelerating applications in the cloud, making them fast and available wherever they reside.
http://resources.
idgenterprise.com/
original/AST-0001239_
F5_Optimized_and_
Accelerated_Cloud.pdf
Securing the Cloudpublished by F5
Cloud computing has become another key resource for IT deployments, but there is still fear of securing applications and data in the cloud. With F5 devices, you can keep your most precious assets safe, no matter where they live.
http://resources.
idgenterprise.com/
original/AST-0001687_
F5securingthecloud.pdf
C o n V e r g e d I n F r A s T r u C T u r e 3 7
resources
03IDGEconv_5assets.indd 35 8/25/11 1:57:24 PM