When it comes to buying a server, the first question to ask is: What exactly do you need? Whether you've made this...
decision already or someone else has tasked you with the assignment, buying a server for your IT department can be a long process. There are a number of different types, models and options, including blade centers, freestanding floor models, 1U rack mounts and more -- not to mention your options if you're considering a virtual or cloud-based environment.
And that's just scratching the surface: There are myriad options to consider, including packaging, functionality, interoperability, performance, availability, survivability, capacity, energy efficiency and the economics associated with servers to navigate.
To guide you through buying a server, we will cover three important components:
- Determining needs and requirements.
- Understanding your available options.
- Putting it all together.
There are many types of physical servers with various features, functions, costs and other attributes, just as there are a diverse set of business and application needs. To find the right server, first understand your business and application requirements. With this information, aligning various server functionalities to meet or exceed these needs becomes easier.
To get started on buying a server, determine what the rest of your IT infrastructure environment looks like. Include how many servers you have, what operating systems and applications are installed and what tools and features you are looking to add.
On the acquisition front, do you typically buy from only one manufacturer, or different ones depending on the specific acquisition or application scenario? Do you have all Hewlett-Packard Co., Dell Inc. or IBM servers, or a mix of them, all running various applications or operating systems? Likewise, is yours an all-Microsoft shop, or do you have a mix of vendors, software and tools?
What does your IT networking environment look like, both locally and for remote or Internet access? For example, if you have 1 Gigabit Ethernet to your existing servers, but you also have a switch that supports 10 GbE, you could include 10 GbE on your new server. On the other hand, if your existing network is not 10 GbE capable, you would have to upgrade those items.
Understanding your networking capabilities is also important if you plan to leverage cloud-based servers, which require networking service in order to access the applications running on them. Likewise, if you're looking to deploy a virtual desktop infrastructure, your network must be able to support the traffic and activity hosted on your new server.
Additional questions to ask include:
How are you currently protecting your IT environment and all associated data? What are your current business continuity and disaster recovery plans, and do you have requirements to enhance your application (and data) availability or survivability via server clustering or other means? Have you started to consolidate or virtualize your application servers; if not, are you planning to?
Do you need to support specialized devices or items attached to your servers? Are there specific requirements for special keyboard, video, mice, USB or serial ports; CD or DVD drives; networking devices; RAID cards; third-party graphics; or specialized adapters?
When buying a server, understand your needs and requirements by mapping IT resources to specific business activities.
With adapter or expansion cards, for example, many new servers come with Peripheral Component Interconnect Express (PCI-E) capabilities, along with some backward-compatible support for PCI-X (the predecessor to PCI-E). The adapters you need for your new server will factor into the packaging or the amount of expansion capabilities necessary.
Do you have a growth- or capacity-forecast plan that reflects business objectives and future demand? A performance and capacity plan can be a simple and straightforward way of showing what you had in the past, including usage or historical trends, as well as current and predicted usage needs. Capacity plans and forecasts can assist in factoring your availability, server response time, storage, networking and software licensing requirements, as well as current and future energy demands.
Could you benefit from tiered servers? There are many different types of servers and a plethora of packaging options, processors and features. Generally speaking, different servers are aligned to different categories, classes or tiers of business and application functionality. For example, the top tier serves the needs of the best in class, mission-critical applications or business functions that require high performance and availability -- and lost time is lost opportunity. The lower tiers of the service focus on cost reduction or avoidance.
As a result, tiered servers refer to the level of service to which various business and application usage is linked. Your organization may have templates, tiers or service categories already defined for how IT resources are used. If not, now would be a good time to consider developing them in conjunction with your solution provider or other resources.
What about the cloud servers? Cloud servers are essentially virtual servers provided as a service from a solution provider on an hourly, daily, weekly, monthly or other on-demand basis. The charges or fees will vary depending on usage, processing capabilities, memory demands, data storage capacity and protection level. For some IT environments, cloud servers can help avoid the cost and overhead of buying and maintaining physical servers and associated software.
While you can run hypervisors on your physical servers and create your own virtual servers -- and even a private cloud -- shifting ownership to a cloud provider may reduce costs even further by driving up the utilization of a shared server.
When deciding on a cloud vs. a private server, take a step back and look at your big-picture business strategy. For smaller environments or special circumstances, moving all or some applications to a cloud, pay-as-you-go or subscription-based model can be advantageous. However, there can be hidden fees and service issues associated with the cloud. Thus, think of cloud servers or services as another tier of IT servers at different costs and availability points.
Bottom line: Start simple. When buying a server, understand your needs and requirements by mapping IT resources to specific business activities. Instead of speaking GHz, gigabyte, ports or protocols to the business people, learn their language and map resource usage to what they understand. For example, prove business value by saying something like, "With X number of servers and Y amount of memory, you can support 1,000 email users each with a specific mailbox size and availability." Approaching both sides of the fence can help you move forward on the momentum of your success.
Greg Schulz is founder of The Server and StorageIO Group, an independent IT industry advisory and consultancy firm based in Stillwater, Minn. He is also author of the books The Green and Virtual Data Center (CRC) and Resilient Storage Networks (Elsevier). Follow him on Twitter at @storageio.