Uncovering the hidden costs of fragmented data

Opinion

Uncovering the hidden costs of fragmented data

CIOs are throwing money away if they do not regularly defragment their virtualised systems, says Thomas Doria, business development manager for Diskeeper EMEA.

Fragmentation is one of the bugbears of modern PCs. It happens when your system has enough storage space or memory to keep and run something, but it's not in a continuous block. So it slows down dramatically every time you use even a relatively simple application, because not only is it running the software, it's having to solve a mini-jigsaw to make the data work every time you type something.

There is an illusion that this problem has been solved. It's often seen as a home user-ish, 1990s Windows problem. However at Diskeeper we get a lot of business from sorting out issues that arise in storage area networks (SAN), network-attached storage (NAS), high-end workstations and other corporate environments which come unstuck when fragmentation sets in.

Disc space

The issue is that computer networks work in very specific ways. Here's what happens: the file system driver, NTFS.sys, assigns the logical location of new data. The physical writing of the data is passed to a fault-tolerant device driver which will typically be hardware or software RAID. This puts the data through a series of policies and procedures and hands it over to the disc device driver.

During this series of processes, information can be "striped" so it's easy for the computer to get at. This is great when you have the right application but less so when the application needs to work differently.

Ironically, the modern day issue of where the data is physically stored is more of a problem than ever. A computer looking for split-up data on its disc is one thing; on a network it's another. On a fully virtualised system it can be a nightmare. The system may end up looking for data split across different continents if the company is using a public cloud infrastructure.

Efficiency issues

This is problematic from a number of angles. First, the efficiency that can be applied to any particular task suffers. Assuming the idea of a computer is to get a task done, taking twice as long to do it isn't good, particularly if multiple users are doing the same thing and so creating bottlenecks in the system.

The problem is exacerbated when you examine the effect on corporate storage. Tests we commissioned from Windows IT Pro confirm that a fragmented disc can appear to be much fuller than it actually is. We found one disc with only a small amount of fragmentation appeared to be 15% empty, when after defragmentation it actually had 22% of its space free.

This translates into extra cost for unnecessary storage, and slow applications, possibly leading to server and network upgrades. One can quickly see how costly fragmentation can become.

And of course it does nothing at all to improve your green credentials or return-on-capital-employed figures for using the extra power and buying new kit.

Not only has fragmentation not gone away, but that it's a key factor in virtualised corporate computing. When their network slows down very few people - CIOs included - think "It's the file system" rather than "It's the computer". But we're willing to bet that, a lot of the time, that's the cause.

Table 1

Virtual environment        
Activity Environment Fragmented (seconds) Defragmented (seconds) Performance improvement
File copy (5Gbytes) Test Case 1 38 31 18.4%
  Test Case 2 80 62 22.5%
  Test Case 3 104 57 45.0%
  Test Case 4 306 61 80.0%
Opening larger Office files (100 pages Word and or sheet Excel 2007 size about 5Mbytes) Test Case 1 13 9 30.5%
  Test Case 2 14 11 21.5%
  Test Case 3 15 11 26.5%
  Test Case 4 25 13 32.0%
Disc-to-disc back-up (using VSS method to the SATA attached discs) Test Case 1 1395 996 28.5%
  Test Case 2 2950 2287 22.5%
  Test Case 3 7200 6600 8.5%
  Test Case 4 11250 11030 2.0%
Anti-virus scan Test Case 1 58 39 32.0%
  Test Case 2 84 59 29.5%
  Test Case 3 105 62 41.0%
  Test Case 4 186 118 36.5%
Bidirectional flows with the Exchange server, server side Test Case 1 8 6 25.0%
  Test Case 2 14 9 35.5%
  Test Case 3 19 12 37.0%
  Test Case 4 26 16 38.5%
Bidirectional flows with the Exchange server,client side Test Case 1 7 6 14.0%
  Test Case 2 14 9 35.5%
  Test Case 3 25 13 32.0%
  Test Case 4 36 19 47.0%
"Bulk" insert into a SQL database of 100,000 random values line Test Case 1 27 22 18.5%
  Test Case 2 34 29 14.5%
  Test Case 3 55 33 40.0%
  Test Case 4 96 49 49.0%
Creation in a SQL database of a series of tables with unique key Test Case 1 17 16 5.5%
  Test Case 2 24 22 8.5%
  Test Case 3 49 43 12.0%
  Test Case 4 76 62 18.5%
Within a complex SQL request Test Case 1 27 26 3.5%
  Test Case 2 29 27 7.0%
  Test Case 3 48 40 16.5%
  Test Case 4 81 64 21.0%
Within a complex SQL request Test Case 1 37 35 5.5%
  Test Case 2 49 37 24.5%
  Test Case 3 68 52 23.5%
  Test Case 4 96 61 36.5%

Table 2

Costs and benefits of defragmenting a virtualised computer environment      
       
Total number of servers 10 50 100
Capital acquisition costs (software licences) £ 2,900.00 £ 13,212.00 £ 26,424.00
Annual software maintenance costs (upgrade assurance, technical support) £ 1,450.00 £ 6,606.00 £ 13,212.00
Initial evaluation and deployment costs £ 1,000.00 £ 1,000.00 £ 1,000.00
Impact to other process/production (including configuration requirements) £ 0.00 £ 0.00 £ 0.00
Operational costs (e.g. IT management reports) £ 500.00 £ 500.00 £ 500.00
       
Total Cost Savings Summary      
Increased system lifespan £ 592.72 £ 2,963.61 £5,927.22
Helpdesk/IT support savings £ 400.00 £ 2,000.00 £4,000.00
Hardware cost avoidance £ 1,208.63 £ 6,043.13 £12,086.25
Energy savings £ 335.06 £ 1,675.30 £3,350.59
Value of improved productivity £ 1,556.06 £ 7,780.30 £15,560.60
Indirect benefits £ 0.00 £ 0.00 £0.00
Estimated total annual savings £ 4,092.47 £ 20,462.33 £40,924.66
Source: Diskeeper    

Email Alerts

Register now to receive ComputerWeekly.com IT-related news, guides and more, delivered to your inbox.
By submitting you agree to receive email from TechTarget and its partners. If you reside outside of the United States, you consent to having your personal data transferred to and processed in the United States. Privacy

This was first published in September 2010

 

COMMENTS powered by Disqus  //  Commenting policy