Replacing vs Extending Server Life

Quite often at Arnclan, when decisions aren’t straightforward we will be asked to produce a report with supporting information on ‘what difference will this make’.

Server replacement being a great case in point…

The report below is an example of the sorts of things we have produced for our customers over the years, so we thought we’d share..

Our Value Add

It’s because we’re faced with these sorts of questions all the time, that we can produce the information much more quickly than customers having to scour the internet and supplier websites and sift through what’s important.

You too can have a report like ours

And as professional advisors, we find it’s part of our service that allows small business IT people to save time as well as money. Both of which are precious commodities to the small business IT support department. Please don’t hesitate if we can assist you with your business case.

Server Replacement v Extended Life.

An Example Report:

This is for information only, and will be particular to the particular organisation at the right time. By the time anyone reads this, it may already be out of date (but hopefully still useful).

Report Objective

Here at Arnclan we believe there is no such thing as a one size fits all solution. We provide IT support to businesses in different vertical markets and of differing sizes.

There are many factors that determine the ‘correct’ solution when considering your company’s IT infrastructure. Typically these could include:

  • Whether business is new start-up or established business
  • Economic climate, cash-flow
  • Future expansion/growth plans
  • Time since last investment in IT infrastructure
  • The need for additional grunt (will like for like replacement be adequate or would faster data processing benefit the business)

Consequently, we feel that as your preferred IT technology partner, or ‘Trusted Advisor’ to use Microsoft’s phraseology, our responsibility is only to arm you with the facts necessary to make an informed decision.

Provided we achieve that then whatever the outcome, hopefully, it will be the right one for you and your business.

Management Overview

This document sets out to provide some information to help determine whether to replace your existing server or to replace just the components (primarily hard disk drive(s) and/or controllers) to eke out the one that you have.

The technology in a typical server isn’t greatly different from that in a desktop or laptop computer. Servers are somewhat more focussed in duty cycle (servers tend to be left on 24/7/365), redundancy (loss of data on a single workstation pales into insignificance when compared to losing the data on a typical workgroup server) and speed.

Like workstations and laptops the performance effecting components are the processor, memory and hard disk sub-system.

Of these it’s the later, the hard disk sub-system, that gives us as the provider of your IT support the greatest cause for concern.

Hard Disk Drives & Controllers

Hard disk drives contain the Windows Operating System and, more importantly, your data files. Until recently all hard disks were functionally similar; a number of rotating platters onto which data is written and read by magnetic heads.

Performance and cost of a hard disk being determined by; reliability (more expensive, ‘professional’ grade, disk drives have better bearings than cheaper, ‘consumer’ grade drives), data access speeds (faster spin speeds result in shortened read/write times) and finally the type and revision of the interface also effects data throughput.

More recently SSD (Solid State Disks) have immerged which are gradually increasing in market share. Currently Solid State Disks are only available in smaller capacities and even here the cost per GB makes them expensive when compared with traditional hard disk drives.

Eventual failure of traditional hard disk drives is certain and to some extent predictable. They do after all have a spindle spinning at, depending upon model, somewhere between 7,200 – 15,000rpm! We would advise against continuing to use any disk that has given 4 years good service in a production environment and closer to 3 years if the drive is one that would be classified as ‘consumer’ grade.

Processors and Memory

Whilst processors and memory modules do fail from time to time, these components will most often be fully functional when a computer, be it server, desktop or laptop, is eventually discarded.

Processors and memory, the consequence of whose failure is a mere inconvenience when compared with the failure of a hard disk, simply get old and become obsolete due to advances in technology.

Advances in Technology

The first small form factor computers were introduced in the late 1970s with the most famous (the IBM PC) arriving in 1981.

Since then, broadly speaking, computer processors, memory have doubled in speed every 18 – 24 months with similar advances in hard disk speeds and capacities.

So a new computer (of similar cost and relative specification) will be approximately 4 times faster than a 3 year old one.

Tech Spec’s & Stats

Hard Disk Drive

Below are bench marking results from disk systems in 3 machines. The machines were:

  1. A fast desktop system (in our office)
  2. Arnclan’s own server (whilst in active use)
  3. Your Old Server (also in active use)

The key figures are ‘Access Time’ (time to access some randomly selected data- the yellow dots tell the story) and ‘Average Transfer Rate’.

Underneath the 3 sets of data is a graph for comparison purposes.

 A fast desktop system

a graph of a fast desktop

 Arnclan baseline server (whilst online in operation)

server baseline graph

 Your Old Server

old server performance graph

 Relative Performance Comparison

server speed comparison graph

Notes:

Random Access Time is measured in milliseconds – faster (shorter time is better)

Average Transfer Rate is measured in Megabytes/second – more (higher throughput is better)

Processor

Many people are aware that processor speed is referred to as ‘clock speed’ measured in GHz (i.e. 1000,000,000 of processor cycles per second). What is not quite so well understood is that whilst GHz can be used to compare processors of similar ‘architecture’, over time fundamental design changes are at least as significant as clock speed.

Since the first IBM PC was launched with its Intel 8080 processor we have seen advances through 8, 16, 32 and more recently 64 bit processing. The consequence of this was, at each change, effectively a doubling in processing power for any given clock speed.

Similarly we have since around 2006 been in the era of multi-core processors. Of which Intel’s Core 2 Duo, Core 2 Quad are examples; respectively further doubling or quadrupling of processing power for no change in clock speed.

Comparison of processor performance is not, therefore, simply a comparison of clock speed, but rather of processing power. Over recent years various benchmarking techniques have been used (mips – Millions of Instructions Per Second was one), nowadays we prefer to use a performance index.

The graph below shows relative processor performance of some current entry level netbook, a Dell Latitude laptop and the latest version of the Fujitsu TX200 server processors compared with that of your current server.

processor speed graph

Advances in technology are such that performance of the processor in your existing server is somewhere between that which you would expect from a netbook and a laptop purchased today.

Whilst the processor in a new entry level server would be over 4 times quicker than that which you are current using.

Memory

The increase in memory transfer rates has been almost linear. This is not really a surprise as the increase in memory speed was driven by the need for memory to keep up with ever faster processors. Below is a graph that shows how entry level memory performance has changed over time.

memory speed graph

Conclusion

The current performance of your server’s main components hard disk, processor and memory is around half of that which you would expect from a £479 Dell Inspiron Laptop!

Whilst replacement of the hard disk drives is something that needs to be done to reduce the likelihood of business disruption and to ensure data integrity you may feel, given the statistics above, that it is more appropriate to consider replacing the server in its entirety.

FAQs

Q: I am currently running Small Business Server 2003, will my current hardware allow me to upgrade to Small Business Server 2008 or Small Business Server 2011?

A: No, the recommended specification for hardware to run SBS 2008 and SBS 2011 is:

 Small Business Server 2011 Standard Edition – System Requirements

Processor Minimum Quad Core 2 GHz (64-bit) or faster

Memory Minimum 8 GB; Recommended 10 GB

Hard Disk Space Minimum 160GB

Your server doesn’t meet the processor or memory requirements for either Small Business Server 2008 or 2011.

Q: If I replace the disk drives in my existing server how much longer could I use it?

A: If nothing else has changed since it was installed then potentially until some other component fails. In reality we most frequently replace servers because a 3rd party vendor (e.g. Sage) release a new version of software and after installation (nobody reads the minimum requirements before installing an update) the system runs really slowly.

Q: If we replaced the server would we have to upgrade to Small Business Server 2008 or Small Business Server 2011.

A: No, there is no requirement to upgrade the Operating System in order to benefit from the enhanced hardware performance. Today’s hardware is currently backwardly compatible with Small Business Server 2003 (although this position will change at some point in the future).

In case you arrived in the middle of the document, the above is an example report that we tend to produce for customers who need supporting information as part of their decision process.

Borrowing the business case from us, as a trusted advisor with no ‘axe to grind’ is part of what helps us to add value to our customers. The above information will likely be out of date by the time you read this

To get your own report, please contact us to discuss. Many thanks!

When is the right time to replace servers?

Everyone wants and needs to extract every last £ of capital investment, that’s the right thing to do. So when is the right time to replace the company server?

When it stops working?

Let’s stop for a moment and think what that means……..

A server is powerful computer, but like all electronic and mechanical equipment at some point it will fail. It isn’t anyone’s fault, not yours, not ours; computers just do that, just like light bulbs blow and old cars, they break down.

You put a call into Arnclan and this time it isn’t just a power cut over the weekend. The server just won’t start.

The realities are:

  • If you don’t replace your server whilst it is still operational you’ll have to do so after it has failed.
  • As the server gets older……the day that it will let you down gets closer.
  • Just like the light bulb, where life expectancy is often written on the packaging, a server’s life expectancy is not a secret; 5 years is typical, but from experience, we’d say closer to 4 years if hard disks aren’t duplicated (with SAS or SCSI).

So what happens when it fails?

Of course, we will sympathise. We always do our best to get hold of replacement parts (most likely available on next day delivery), reload the operating system (about half a day’s work) reinstall your business application software (say, another half a day’s work) and finally restore backup data (which, of course, could be from the night before the failure).

It is an emotional time, we’ve been through it with new customers many times. It is worth repeating the obvious, because you definitely don’t want to be in this position

If your server isn’t replaced whilst still operational you’ll have to do so after it has failed.

Can this scenario be avoided?

Yes, absolutely it can. The scenario described is the inevitable outcome of not making a conscious decision to replace a server before it fails.

Adopt a strategy to replace ageing equipment before it fails in service.

On the one hand, in the current economic climate, you want to squeeze every last bang out of your hardware buck. On the other, the cost of downtime (potentially 3 days of it) that will result from a server failure could have a dramatic effect on your company’s bottom line.

So in the absence of a crystal ball, when is the right time to pull the plug on the company’s existing server?

The answer is server life-cycle management

There are many different approaches to managing hardware life cycle, taking account of company size, number of offices, financial priorities, but the most import thing about a life-cycle plan is to have one!

Many large corporations, for example, make decisions about how long computer hardware will last at the time of purchase. By specifying this from the beginning these companies are able to:

  • Budget for acquiring new systems at that time
  • Plan to evaluate the systems prior to that time
  • Analyse what condition they’re in
  • If necessary, for financial reasons, take steps to extend their useful life by replacing or upgrading components

In the absence of an existing life-cycle plan we suggest that 4 years is a reasonable life expectancy for your server. To run it past 4 years without replacing, at least, the hard disks, your company runs an ever increasing risk of inevitable server failure.

We would very much like to help you avoid this scenario

Hence this post, because we never enjoy these situations either.

The very last thing people think about is prevention, urgent things have a habit of getting in the way of the important. But why not let us do all the planning for you? That’s what we’re here for.

Drop us a note and we’ll be delighted to put forward a range of options. If you don’t do it now…..

New Website

Feel free to drop us a note whilst we get some content together!