by November 1, 2004 0 comments



If you have just two or three PCs in your office, then you can easily share data between them by sharing folders. This sort of sharing is known as peer-to-peer networking. But it becomes extremely difficult to manage when you have more PCs. That’s when you need to put a server in place. In this article we will give you an overview on choosing the right server for your needs and answer the questions you are expected to face while buying a server. Before deciding which server to buy, you must understand what it can be used for. The following are the basic applications of a server:

Workgroup file and print: This server is used to centrally store data of users and to share things like printers. The load on such servers is not very heavy and you can use lower-end servers to do the job. What is required for these servers are faster hard disk drives and possibly more cache on the storage controllers and disk drives. CPU speed and number of CPUs don’t matter so much here.

Application servers: These servers run applications such as e-mail, Web server, database and accounting. For such servers you require a faster CPU, or possibly two, and huge amount of memory. Fast disks are also required if databases come into picture. Depending upon the mission criticality of the application, you may want to have high degree of reliability. Reliability can be achieved by having redundant (hot spare) components within the server or by having a separate hot-spare server.

Vendors

Branded 
HP-Proliant series
Dell-PowerEdge series
HCL-Infiniti Global Line
Acer-Altos series
Wipro-NetPower series
Zenith-Premium Data/Net server
IBM-xSeries
Sun-SunFire series
Others

Connoisseur
Edge
RDG
Global Infotech
SpiceNet

Infrastructure servers: This is the machine that runs your firewall, Internet gateway, proxy and the like. It is similar to the workgroup server. But, you may not want the fastest disk drives or processors for such a server. Many organizations also use one or more desktop PCs to do this job.

You can easily use a desktop class machine as a server, but don’t use it if you can’t afford to have the server go down. Use a server class machine for that. There are a lot of differences between a server and a desktop machine. The server is designed with a lot of redundant components so that it doesn’t crash all that easily, which is also the reason that it’s more expensive than a regular desktop machine. Also, fitting a desktop PC for a server role, although quite luring in the beginning, may not be a good decision in the long run. You may save on some money initially, but the performance penalties may be too much to overshadow the money saved. Plus, you may never get the kind of reliability and manageability that a server provides.

Server hardware
Let’s look at how to choose the right hardware configuration for your server. Usually an Intel x86 architecture-based server is a good option for workgroup requirements, whether for file or applications services. Let’s discuss various elements of a server in detail. 

Memory
Memory, like number of processors, is based on the applications and user load on the system. These days, when 256 MB is becoming standard on the desktop, 512 MB is the bare minimum for a basic workgroup server. For more intensive tasks such as, messaging, databases, ERP, Web server, you may consider buying even more memory.

When it comes to servers, the amount of memory is not the only decision to be taken. The type of memory is equally important. For heavy workloads, where large amounts of data have to move, to and from the memory, registered memory can give a big performance boost. Also for applications such as dynamic Web serving and e-commerce, frequency (or the speed) of the memory is also very important.

Processors
For a multi-processor machine, Intel Xeon and AMD Athlon MP based servers are available. For single-processor, entry-level servers, the P4-based servers are there that are more cost effective. If you take a server that supports two processors, then it’s better to take it with both the processors. Many times, people tend to take a dual-processor server, but buy only a single processor in it. This is not recommended because processor lifecycles are short, and by the time you’re ready to add a second one, it may not be available in the market. 

The choice of the number of processors depends on the specific applications you will be running on the server and the number of users that will have to be supported. The standard choices are single processor, dual processor and four way. Of course, there are servers that can take more processors, but for the business size that we are considering they would not be required.

Storage
At one time, SCSI was the only type of hard disk that could be put on a server. Today’s IDE disks are capable of taking entry-level server-class loads. Serial-ATA disks are also fast becoming big contenders to SCSI, at least in terms of performance if not in reliability of data.

But, choosing server storage is not just a question of choosing the type of disk and the capacity. You could opt for disks on the server, disk array, separate from the server, or a separate storage device called a NAS (Network Attached Storage) device. If you choose direct attached storage (disks on the server), then it is advisable that you opt for a RAID (Redundant Array of Inexpensive Disks) setup rather than having just one or two disks. Simply put, RAID combines multiple disks to perform as one, improving reliability and data throughput in the process.

There are various types of RAID, known by levels, from RAID zero to RAID 5. You would normally not need RAID 5. 

Networking
Servers will have network cards built in. Gigabit LAN has become standard on servers. But, if your network is not Gigabit, you may not be able to attain benefits out of it. But then you can have a Gigabit backbone connecting your servers, leaving rest of the network in its present 10/100 Mbps state. This will improve end-user experience of the server, as the network pipe connecting the server is no longer a bottleneck. If your application requires even more data transfer, then you may get better performance by having more than one network card. The trick is to put half the users on one card and the other half on to the other.

Chassis, power supply, cooling
This becomes important only if you are buying an assembled server. A server tends to become much more hotter than a normal PC. So there should be enough space inside the chassis for air to circulate and for the heat sinks and fans. Normally, you would require a fan, not only for the CPU but also for the disk drives and the chassis, in order to cool the server components properly. 

Power supply is a critical component for a server. It should be able to take the load of not only the components currently installed in the server, but also of any components that you may add in the future, such as additional disks. For more reliable server operation, you may also go in for dual power supply, where one acts as hot spare.

Redundancy
Needless to say, a server is much more critical to your business than any PC. To ensure that the server is available, a high degree of redundancy can be built into it. This starts with a standby power supply, all the way to a standby RAM, disks and CPU.

In addition, the redundant parts are normally plug and play. That is, you can pull out a failed one and replace it, even while the server is running. When choosing a server for mission-critical applications such as ERP, keep an eye on redundancy features.

Form factor
Gone are the days when servers like desktops came in one-size-fits-all configurations. Today you can choose from normal servers, rack servers and blade servers. Blade servers are for organizations that have a very large number of servers and want to fit them into a very small space. So we are not discussing them here. If you are likely to end up with, say, ten or more servers in one location, it may be worth your while to consider rack servers. These, as the name suggests, are servers that are made to look like racks in a cabinet, and actually fit in to a cabinet, making it easier to manage as well as saving on space. For most other requirements, pedestal servers are just fine.

Expandability
Servers, like PCs do offer expandability options. In fact, expandability is required more in a server than in a PC. But you are unlikely to make full use of all of them. When the situation comes you are more likely to buy a new server than expand the existing one, unless you are talking about a very high-end server. But basic expandability, like the capability to take more RAM, an extra network card and a few extra hard disks should be there.

Manageability
Managing a server can become both a complex and expensive affair, if not done properly or if right components are not chosen. If you want the ability to remotely manage a server, not just monitor, you should seriously consider choosing IPMI compliant manageability hardware. Such hardware may be built-in or the server may have a slot to add it. It lets you control the server irrespective of the OS state, using any IPMI compliant management software, also provided along with the hardware. This means is that you can be in full charge of the server, remotely, as long as the power cable is plugged in, no matter whether it is booted into an OS or not. You can remotely diagnose the server, switch it on or off, get alert messages and so on. ¨

Q&A




I have just bought a new server, but the performance is not very good. What could be the reason? 
There can be many reasons for this, as it’s a fairly common problem. It depends upon the application you’re running on the server. If you’re using it as a proxy server for instance, then the amount of RAM you have, could be an issue. If it’s a database server, then possibly the hard disk needs to be checked out. Many times, a single server is used for running multiple applications. This could overload the server; so keep a watch on the CPU and RAM utilization on the server. Also check which services are consuming the maximum resources. You can often improve the performance of a server by turning off unnecessary services from the server. 

My server cannot take the current load, should I upgrade it or buy a new one?
Unlike desktops, server upgradation is not an easy task and the cost could be very high. Instead, it is advisable to buy another server and distribute the load between the two. This will guarantee at least some minimum amount of resources, available from the two servers to your applications.

However, in some specific cases, you may not need to buy an extra server and minor upgrades may make your applications run faster. But, for that you need to identify the nature of your applications. For example, if you are running a file server then adding a faster hard disk or a disk controller with more amount of cache or a RAID-5 configured disk array can dramatically improve file-serving performance. But the same may not help a Web server, because there the requirement is not always faster storage, but faster CPU and more RAM. So, identify your applications and their requirements carefully and then take a decision.

I am buying a server, how many resources should I buy upfront, and how much room should I leave for upgradation?
When buying a server, it is advisable to buy critical components, such as CPU, Power supply and RAM, upfront. That is, if your present needs are for a single CPU, but would like to add another CPU afterwards, it is better to buy two processors right now, because later on processor availability and pricing could become an issue. You can leave room for things, such as disk drives and network cards.

I have bought a new server, what do I do with the old one?
Move it to less intensive tasks, such as proxy and firewall or make it a standby or backup server.

What is pre-failure warranty?
Pre-failure warranty covers critical components like memory before they actually fail, making your server operation more reliable. In IDE hard drives, there’s a feature called S.M.A.R.T, which is a set of tools built into the hard drive itself. They keep monitoring the hard drive for certain critical parameters and warn you in advance if any of them is going bad. A SMART message usually indicates that your hard drive is going to crash. Back up all your data and have the hard drive checked immediately if this is the case.

No Comments so far

Jump into a conversation

No Comments Yet!

You can be the one to start a conversation.

<