Gartner may have scratched its Spring IT Expo, but AFCOM's Data Center World 2009 is on. The convention is March 8-12 at the Paris Hotel and Convention Center in Vegas. The major theme this year -- how to be green in lean times...
<script type="text/javascript" charset="utf-8" language="javascript" src="http://static.polldaddy.com/p/1302947.js"></script> Which flash-based storage product has the most promise for your enterprise?
( surveys)
<script type="text/javascript" charset="utf-8" language="javascript" src="http://static.polldaddy.com/p/1302926.js"></script> Which technology do you think is best for converging data and storage networks?
( polls)
<script type="text/javascript" charset="utf-8" language="javascript" src="http://static.polldaddy.com/p/1302967.js"></script> Does pairing Web services and storage provisioning make sense to you?
( polls)
<script type="text/javascript" language="javascript" src="http://static.polldaddy.com/p/1164603.js"></script> Which technology will be the hottest this year?
( polls)
<script type="text/javascript" language="javascript" src="http://static.polldaddy.com/p/1094458.js"></script> Which of our 10 start-up picks has the most promise for your enterprise?
( polls)
Participate in this poll and let us know whether you’re feeling the affects of changes to IT budgeting and spending.
<script type="text/javascript" language="javascript" src="http://static.polldaddy.com/p/1115470.js"></script> Has an IT budget crunch affected your work?
( polls)
Credit Suisse was a very early server virtualization devotee, as I mentioned in my earlier post. That means virtualization comes into play at just about every layer of the IT infrastructure. Next up, quite possibly, is memory, said Steve Yatko, managing director of Credit Suisse's Global Research and Development Group, in a recent interview. Ideally, Credit Suisse would like to uncouple the processor and memory so that they can scale for an application workload independently. This is particularly important in development and testing, which could really benefit from an unprecedented amount of consolidation on a physical box and that independence of scale, he said.
"Memory is probably going to be one of the last key areas that we tackle as an underutilized resource as we go after a modular compute fabric. So we basically would have compute, memory and I/O bricks all scaling independently in a modular way and building a new system that is a composite architecture."
For more on how virtualization is coloring enterprise IT, go to our special New Data Center package.
My continuing series showcasing virtualization insight from virtualization pros today focuses in on Credit Suisse.
This global financial services firm is one of those companies that recognized early on that virtualization would be transformational for its business. By 2002, IT managers understood that virtualization wasn't just a core technology, said Steve Yatko, managing director of Credit Suisse's Global Research and Development Group, in a recent interview. "We started to see that virtualization was beginning to build more of an ecosystem and a landscape shift across many different facets of technology and business capability. We saw it transcending into a paradigm shift allowing us to enter what we would consider a whole new era of computing, namely service-oriented computing."
With that understanding, Credit Suisse approaches virtualization not only as the layer that allows for the orchestration of capacity across servers and desktops, storage, memory, I/O and networking in general, he says, but also as a way to enable a virtual workforce, virtual applications and secure virtual access for internal and external users.
Essentially, Yatko said, Credit Suisse approached virtualization in a services model. "The goal was to leverage virtualization to build a much higher level of abstraction from the bits and bytes, the LUNs, the packet sizes, into something that could really talk about the business capabilities users needed to achieve and how much money they were essentially willing to afford for those capabilities."
Read about Yatko's must-have virtualization tool choices.
Get more information on how virtualization is coloring your world in our special New Data Center report. Read more
I recently interviewed more than a dozen enterprise IT executives about their must-have tools for virtualization (view a slideshow of some of their choices). Beyond the actual tool choices, they had lots of interesting things to say about how they came to embrace virtualization and how they're managing the virtual infrastructure so far. I'll be sharing some of that insight over the next several blogs, starting here with the perspective of Syed Azhar, vice president of IT operations for Plug and Play Tech Center, a Silicon Valley-based community of more than 120 technology start-up companies. Plug and Play Tech Center "takes the plumbing out of the start-ups' hands so they can take care of their core competencies," Azhar describes. Virtualization, for its flexibility among other reasons, is a natural fit. Plug and Play can easily spin up virtual servers on demand for its clients, for example. Interestingly, when asked to select a choice virtualization tool, Azhar names a managed service, not software or an appliance. Plug and Play relies on NetMagic's Remote Infrastructure Management services to monitor and manage its infrastructure on a 24-hour basis. Plug and Play gets the comfort of knowing somebody is always watching over the virtual infrastructure and, in turn, can offer its start-up clients strong service-level agreements, Azhar says. "This is absolutely a business must-have – either we do this or we hire a bunch of people" -- which he says the company can't cost-justify at its current size.
For more virtualization coverage, see our special New Data Center package, The virtual spectrum: how virtualization is coloring your world.
At the Pacific Northwest National Laboratory, in Richland, Wash., “lean provisioning” has turned into a virtualization must-have, said Daryl Anderson, portfolio manager for application and data hosting/housing at the lab, in a recent interview. Lean provisioning is similar to thin provisioning, an increasingly popular feature of storage virtualization that allows automated resource provisioning across a storage area network. This U.S. Department of Energy lab has been using lean provisioning, its own version of thin provisioning made possible by the LSI storage appliance it uses, for more than a half a dozen years. While this process is more manual than today’s thin provisioning, PNNL is still realizing savings, Anderson says. Read more
Cisco wants in on the data center -- the virtual kind, that is. If that message has failed to escape your notice till now, then the company hopes its latest raft of announcements will drive home the point. With the Cisco Live! customer conference as the backdrop, the company revealed more of how it intends to participate in the virtual infrastructure. VFrame, its network-driven service orchestration provisioning platform, remains a centerpiece. An enhanced version, VFrame 1.2, will support integration between Cisco's Application Control Engine (ACE) software and VMware ESX virtual servers. VFrame 1.2 could virtualize servers to ACE virtual devices, for example.
Read more on Cisco's Data Center 3.0 plans
Read more on how Cisco plans to manage the virtual data center
Listen to blogger Mitchell Ashley on VFrame's significance
Jim Metzler and Steve Taylor are two of our resident WAN optimization experts, frequently sharing their insight in the Wide Area Networking Alert newsletter. In a recent conversation, the dynamic duo shared some great insight on where we are on the long road toward end-to-end optimization. To Metzler’s way of thinking, we’re moving from Stage 1 to Stage 2 of network optimization. In Stage 1, he says, the idea that IT should be assuring acceptable application performance reached the widespread enterprise consciousness. As such, IT undertook tactical approaches, solving problems on a piecemeal basis using WAN optimization controllers and the like. Now comes the hard part, and there are no simple answers in this stage. Enterprises have got to figure out how to assure application uptime, performance, management and security across myriad organizational and technology boundaries. Simply put, “We don’t do a good job of solving problems that cut across organizational boundaries.” Taylor agrees, of course. He puts it in the context of convergence, which we’ve been talking about for years. “When you really can’t tell the difference between the application and the network then we truly have convergence. And we’re only about 20% of the way there.” Read more
Read more about how net optimization is evolving.
Little by little, management tool vendors are getting better at providing the ability to manage the virtual server infrastructure. HP, for example, has renewed efforts to develop software for managing VMware hypervisor technology. In partnership with VMware, it will tackle management and automation software for performance, configuration and availability management of VMware servers. (HP already handles basic VMware management.) And start-ups such as ManageIQ are focusing on capabilities such as virtual server configuration management. Truly, these efforts are good. But they still fall short of the ultimate management need – meaning, the ability to provide end-to-end management of the entire virtual infrastructure, not just the server piece of it. As Cameron Haight, research vice president at Gartner, says in a recent story on the challenges of virtualization management, “It's important to look at virtualization in a holistic fashion [because] poor design in one IT silo can impact the overall performance. It's important to have management visibility across these technology components to help us rapidly diagnose potential performance and availability problems.”
Read “Virtual goo” for more expert insight on how to handle the challenge of managing and optimizing the virtual infrastructure.
As technologists, the last things you probably want to deal with are people and processes. But if you’re trying to optimize application delivery, then an organizational overhaul most likely will be required. Michael Morris, a network architect and frequent Network World blogger recently described the issue for me:
“A lot of IT shops are set up in the old silo model -- here’s the application piece, here’s the infrastructure team and underneath the infrastructure team I’ve got the server team, the network team, the security team and they’re all in silos under different managers. It’s very hard right now to say, ‘OK, I’ve got a new application coming in. What do we need to do to deliver this application so it’s optimized on the network, has the right security, the right server resources, and all of that goes in harmony?’ IT organizations aren’t built that way.
“You get a new application, like SAP, and it’s not until six months into an eight-month project that the developers come to the network guys and say, ‘Oh by the way, we’re going to run SAP over your network. Is that OK?’”
The application delivery story, Morris says, is really a non-technical one. It’s about how enterprise IT needs to organize for proper service delivery optimization and what processes need to be put in place to make sure that happens.
That might not get you super excited, but it’s the right move – and your users will love you for it.
For more on this topic, read:
Dear IT: Forget the technology
Four ways to improve IT and get better application performance
Cisco would like to control the virtual infrastructure via products that automatically provision, manage and optimize virtual environments. Through its VFrame data center appliance, for example, it proposes a way of pulling together not only the network capacity needed to meet application demand but also the server and storage resources. At least one industry observer, Zeus Kerravala, a senior vice president at Yankee Group, finds intrigue in the company’s strategy – especially since it goes against what other IT vendors – meaning, those with a computing bent -- say. “The network does touch everything, so it makes sense that eventually the network could be the orchestrator for virtual environments,” he says.
Plus, Virtual goo, how to manage and optimize the virtual infrastructure
In a data center strategies briefing earlier this year, Burton Group analysts shared their thoughts on technology’s biggest up-and-coming trends. On the storage front, one of the trends noted by analyst Gene Ruth was the growing acceptance of backup as a service. Storage providers are finally shaking the stigma they’ve born since the dot-com era, he says. EMC’s purchase of Mozy and IBM’s of Arsenal Digital helped a lot to give legitimacy to the storage-as-a-service market. Too, as discussed in this week’s special New Data Center package, storage service provider wannabes have smartened up. They don’t want your primary data, as their brethren of old did. They want to store your backups and e-mail archives, and offer the opportunity to relinquish your worries over the backup infrastructure and associated IT staffing. Many IT managers are digging the idea, are you?
Read the story, Is an online backup service OK for your data stockpile? Read more
See who's offering storage-as-a-service
See the New Data Center package in its entirety
... or at least that's what EMC is hoping for its enterprise-class flash drive technology. CEO Joe Tucci touted its Enterprise Flash Drive (EFD) technology at EMC World this week, even saying the advent of flash technology will change the storage industry more than anything else will over the next 10 years. And, in a Network World Panorama podcast posted this week, Barry Burke, chief strategy officer of the Symmetrix line, says the interest in enterprise-class flash "is far broader than we expected" because customers are considered it for "just about any type of application in their environment that you can imagine." This, he says, kind of threw EMC for a loop, as it had initially considered EFD primarily -- if not exclusively -- for high-performance, mission-critical, heavy transaction-processing-oriented applications. But performance improvements are performance improvements, it seems, and customers are seeing no reason not to consider EFD for applications like e-mail.
Listen now for more details on EMC's enterprise-class flash technology, performance improvements, cost, efficiencies and more.
The notion of storing data in an Internet cloud should be pretty exciting for any Web business – especially for any Web business that isn’t also a storage expert. That’s because cloud-storage providers can relieve the storage burden while speeding data delivery for a Web site operator. Cloud-storage providers – meaning, for now, Amazon, Nirvanix and Mosso -- balance server loads and move data among various data centers to ensure that information is stored close to where it is used. You might have your typical security and reliability concerns, but that’s the old risk-reward analysis.
Read more on storage cloud services.
Listen to a podcast explaining the relationship between cloud and utility computing.
Get more analysis on the latest in storage.
By now we’ve all heard about the looming data-center power problem. But do we truly understand just what it we’re facing? Perhaps not really, contends Ken Brill, executive director of the Uptime Institute, a widely respected data-center consulting firm that has been tracking data-center energy consumption for years. “I’ve been ignoring [energy-consumption growth rate data] as somewhat alarmist – until we got in this recent Uptime member data,” says Brill in a video cast on the group’s findings. Now, he’s really taking notice. Recent statistics gathered from Uptime member companies, representing about 50 data centers, absolutely show an unprecedented growth rate in power consumption, he says. While the average data-center site shows an 11% compound annual growth rate in 2006 and 2007, the top third of data-center sites studied show about a 24% CAGR in 2006 and 2007. In two years, energy consumption has more than doubled at these sites. Data-center electric consumption is the fastest growing sector in the U.S. economy, and the U.S. doesn’t have the means of licensing as many power plants as will be needed to keep pace with these consumptions rates, Brill says. As Uptime keeps crunching the numbers, it promises to share more insight on the implications for the nation’s IT infrastructure. In the meantime, listen to Brill’s current analysis here. And, for more on the data-center energy consumption issue, check out:
Power: What you don’t know will hurt you
Power: the cost reality
5 ways to curb data-center energy use
Energy-efficiency case studies
Our Data Center research center will keep you up to date with data-center news and analysis.
The Leader in Network Knowledge
