Tag Archives: James Staten

Is it Time to Toss Your Servers?

our beautiful site

Abaca Technology Corp., which launched in 2005, offers physical and virtual anti-spam appliances. The company uses Amazon’s EC2 cloud computing service to host the software that works with its appliances. Without EC2, it would have been much harder — and much more expensive — to launch Abaca, according to Bill Kasje, vice president of Business Development. “As a small company, we were able to get our servers up and running quickly,” he says. “We didn’t have to invest in a big infrastructure environment, or have backup power and redundancy, all the things our customers expect from us because email is a mission-critical application.” Though Abaca does deploy in-house servers, it would need at least four more, in a cluster configuration, if it were hosting its software in-house, he says. This way, Abaca’s IT team could focus on the company’s core competency: filtering spam. For many small companies, the smorgasbord of newly available off-site or “cloud” computing offerings means they can reduce the number of servers they purchase and maintain in-house. In fact, according to James Staten, principal analyst as Forrester Research, they may no longer need servers at all. “There are now options available over the Internet that didn’t exist before,” he explains. For instance, companies used to use servers for file sharing, but there are many Internet-based options such as Microsoft Windows Live, Dropbox, and so on, that provide the same options over the Web. “Backups can be done over the Web too,” Staten adds, “and it looks almost exactly the same as when you back up to a server. A lot of people have print servers, but that’s not really necessary any more either, with today’s network-based printers, or wireless-enabled printers with built-in print servers.” In fact, Staten believes, many small companies no longer need any in-house servers at all. Better without servers Why reduce or eliminate servers? “The number one advantage is it gets you out of the IT business,” Staten says. “You don’t have to worry about high availability.” [A high availability configuration ensures continued function by connecting two or more servers in a cluster so that one can “fail over” to the other in case of a problem.] You no longer need to worry about off-site backups, emergency power supplies, or how your company would preserver its data in a widespread disaster like a hurricane, since all these protections are now provided by an off-site by a service provider, and defined in your contract. You can also get by with fewer IT staff, Kasje says. “We would have to have IT staff monitoring systems around the clock. All we have to deal with are the software issues, so that’s much easier. There is a whole class of problems we don’t have to address.” Not having servers on site means much lower upfront costs, though it also means ongoing costs to pay for a service or server space. “You’re trading capital expense for operating expense that you can adjust up or down, depending on your needs,” Staten says. And while the day-to-day costs may be similar, or perhaps lower for owned equipment amortized over several years, off-site servers can provide lower cost if you take risk into account. “There are so many more things to account for,” he says. Three off-site options For companies that want to cut their server count and turn to Web-hosted options instead, there are three different basic options to choose from: Software-as-a-Service (SaaS) In this approach, an application is provided by a SaaS provider and runs on its servers. Your employees (or customers) use the Internet to log into the software. Well-known examples include Salesforce and Google Documents. Hosted servers In this approach, you contract for server space — or even an entire (real or virtual) server at your provider. In many ways, you can treat this off-site server as if it were a regular server, loading applications and data onto it as you see fit. However, the hosting provider maintains the server, usually providing backups, security protections and such. Rackspace and Hostway are two examples of this approach. Raw cloud space “Cloud” is a relatively new term that is often used to describe any Web-hosted offering. Strictly speaking, it simply refers to the architecture by which software and/or data reside in a network or “cloud” of servers connected by the Internet, rather than on a single machine. You can lease raw space in the cloud, for instance, from Amazon’s EC2 service. In this setup, you are still responsible for managing your own server space. If your provider had an outage, in the case of SaaS, the application would be up and running as before once the outage was over. In a hosted server setting, the provider would restore data on the servers, providing the configuration you had before the outage. In a cloud computing outage, once the outage was over, your IT staff would need to reconfigure and reload your online server with the applications and data that were there before. You would be responsible for ensuring backups, and also the security of your data. Because of these added tasks, Staten doesn’t recommend pure cloud computing for small companies unless they also have solid in-house IT expertise. On the other hand, he says, “If you’re really tech savvy, these are great new options to avoid ever having a server within your walls.” Whichever option you choose, Abaca’s Kasje recommends giving off-site computing a try. “You can step into this very easily,” he says. “And you should be able to figure out very quickly whether it’s something that can benefit your business.”

Before You Virtualize Servers: A Checklist

our beautiful site

In the Fall of 2007, the IT department at Campbell Clinic Orthopaedics faced a problem: several servers were due or overdue for replacement, so the clinic would have to purchase up to 10 new ones during the coming year. “We could have gone the traditional route and bought 10 servers,” says Justin Lauer, director of IT. Instead, the clinic bought three physical servers and installed virtualization software on both new and existing machines. Today, Campbell Clinic has 20 to 25 virtual servers running at any given time. The number varies because Lauer and his team create extra virtual servers as testing environments for new software, and then delete them when they’re through. Virtualization has worked out great, he says. “We did it because of an aging infrastructure, but we all got the benefits of consolidation and the flexibility to create new servers at will. And we didn’t realize until we did it how good it would be for our disaster recovery plan.” Since a virtual server is stored as a single file, it’s easy to reload that file on new hardware and get it up and running in case of need, he explains. But while virtualization brings major benefits, it also comes with a few challenges that should be addressed right from the start. Before you virtualize, make sure you have answered the following questions: 1. Will our servers need more RAM? The answer is almost certainly yes. “RAM is the key. I can’t stress that enough,” Lauer says. “RAM is the most used resource in a virtual environment, so don’t expect to get a server and throw a couple of gigabytes of RAM in it. You’re going to max out the RAM. It’s common for servers hosting virtualization to have 32 gigabytes.” 2. What about other resources? For x86-level servers (the low-end, desktop-computer-like servers most small companies use), virtualization can strain more than just RAM. “You have to plan for all the different hardware requirements,” Lauer says. In addition to RAM, these could include more processing power, network input/output (I/O) capacity, and bandwidth. “If you’re trying to run 10 virtual servers on a physical server with only two network cards, that will create a bottleneck,” he says. 3. Are we prepared to cool “hotspots”? One of the best arguments for virtualization is reduced power consumption: fewer physical servers will draw less power, and also require less cooling. While all this is true, your newly virtualized servers will be working much harder than your servers did before, and this can create isolated “hotspots” where temperatures can rise rapidly. “The amount of heat you need to dissipate can increase by factors of 10, especially if you’re putting in blade servers or high availability hardware,” says Bob Waldie, CEO of Opengear, which provides infrastructure management. “You were using hundreds of watts, and now you’re using kilowatts on that one device.” 4. How will we deal with hardware failures? “When you had a larger number of servers running one or two applications each, a hardware failure would only have removed a little piece of your functionality,” Waldie says. “When you have eight or 10 servers running on one physical device, you could lose a lot more mission-critical function if the hardware goes down.” This is why many smart companies put in high availability solutions, with failovers in place in case a server does down, at the same time that they virtualize. 5. Can IT staff deal with increased complexity? “If you had 25 servers before, and now you have far fewer, it may seem like you’ve reduced the complexity,” Waldie says. “In fact, you’ve increased it. You still have exactly the same number of servers running, only some of them are now virtual. In addition, you have the virtualization software itself to manage.” How will your team cope? Training is the best strategy, according to James Staten, principal analyst, Forrester Research. “Make sure whoever will be responsible for the virtualization software is certified on that product,” he says. “If not, you should send them for certification.” 6. Are we ready for the future? Waldie advises reviewing the physical environment and architecture you’re currently planning to use for virtualization, with a view to the next three years. Will your racks and other equipment be able to handle the densities, workload and cooling needs you’re likely to encounter, especially if your company is growing? “Even though adoption of virtualization is ramping, the tools don’t generally work across platforms,” he says. Yet mergers or other developments may force you to work with different brands of virtualization software. “You should get some expert advice about how to architect your virtualized system to gain flexibility,” he says. “It seems simple, but it’s not.”  In time, it will probably get easier to work with two or more brands of virtualization, and “platform-agnostic” tools will be developed, he says. “The thing about virtualization is, it’s still very early days.” SIDEBAR: Want to Try Out Virtualization Software? To truly reap the benefits of virtualization, you’ll want to install a hypervisor that replaces the operating system as the “bottom-of-the-stack” on your server, and runs everything in a virtual environment. If you need to quickly add a virtual server to an existing setup, or just want to give virtualization a try to see how it works, other software options will allow you to add one or more virtual servers to an existing setup. Many providers offer low-end versions of both types of software as a free download. Here are just a few: Citrix XenServer: Built on the open-source Xen standard. Microsoft Hyper-V: Released in June, works with Windows Server 2008. VMware Server lets you add one or more virtual servers to an existing server. VMware ESXi is a hypervisor.