Dear Network Administrators,

This CPTTM NetAdmin newsletter is to bring useful news to you, network administrators in Macau, for references without obligations, so that you can do your jobs easier and better! Hope you like it. if you'd like to unsubscribe or recommend your friends to subscribe, just email me at kent@cpttm.org.mo. Old issues are available here.

Topics in this issue:

• Advancing your career as a network/system administrator
  
• Be careful with RAID5 and SATA disks
• Enterprise router at 1/5 of the cost of Cisco's?
• Upcoming courses for network administrators

Advancing your career as a network/system administrator

Most of the time we are swamped in daily issues like fixing problems and etc. But have you considered the bigger picture of how to advance your career? For example, in five years from now, will you be still in the same post doing the work as today? Do you want to become a senior network/system administrator, DBA, help desk manager, information security officer, IT manager, or even IT director or CIO?

You may think these are impossible, but there are actual things that you can do to increase your chances. You first need to see what the gap between your current status and the desired position is. Usually there are several types of gaps:

• Gap in the breadth of skill. For example, you may be good at Windows server but knows nothing about Exchange or Linux. The easy solution is to write down a simple plan, like, "I will attend a course on Linux in 2010 Q2" and then stick to it.
• Gap in the depth of skill. For example, you aren't strong enough in Windows server. This gap is much toucher to close. Typically training cannot turn an average cook into a good cook. A way to close this gap is to keep practicing (keep cooking and adjusting for better results) and learn from experts (observe how great cooks do it). A great way to practice is to attempt to answer questions on technical forums and verify the solutions offered by others. For example, commit to attempting one question every week.
• Gap in aptitude. This is the hard part. If you're used to blaming the tools, your boss, your colleagues or Microsoft instead of solving the problems, then it is difficult to advance to a senior post.

Below are the major skills that I think should be important for each post:

Be careful with RAID5 and SATA disks

If you're using or going to buy SATA disks to be used by a server, you should be careful in deciding to use RAID5. This is because a SATA disk will encounter an error reading a sector once it has read a certain number of bits. What's the problem with RAID5 then? If one of the disk in a 3-disk RAID5 array fails, once you plugin a replacement hard disk, the two disks must be read from start to end to reconstruct the array. If one of the disk has reached that limit and encounters a reading error, the array will fail to reconstruct.

How likely is it? Many SATA disk has an bit error rate of 10¹⁴. It means that the disk will be unable to read a sector if it has read 10¹⁴ bits. If the disk is say 1TB in size (10¹²) which is 8*10¹² bits. If you read the whole disk from start to end once, there is a chance of 8*10¹²/10¹⁴=8% that it will be unable to read a sector. During the reconstruction phase, the probably of having exactly one disk failing is  2*8%*92%=14.72%. The actual probably for failure is higher because the probability of both disks failing was not included.

In addition, the larger the disk is, or, the more disks you have in the array, the more likely the reconstruction will fail. For example, if you have six disks and each disk is 2TB, the probability for exactly one of the remaining disks failing during the process is 5*16%*(84%)⁴=39.8%. Again, the probability of multiple disks failing was not included. It means that once a disk is replaced, RAID5 may be degenerated into no RAID at all.

So, what should you to prevent this problem? There are several solutions that you may consider:

1. Use RAID6 instead of RAID5. RAID6 uses two disks to store the parity. So, even if a disk can't read a sector during reconstruction, it can still go ahead. The probability for exactly two of the remaining disks failing during the process is (5*4/2)*(16%)²*(84%)³=15.2%.
2. Buy SATA disks that have a lower bit error rate. There are some offerring 10¹⁵ which is ten times better.  The probability for exactly one of the remaining disks failing during the process is 5*1.6%*(98.4%)⁴=7.37%.
3. You could buy SCSI disks that achieve a bit error rate of 10¹⁶. Of course, they are also much more expensive.

Enterprise router at 1/5 of the cost of Cisco's?

We all know that Cisco routers are very expensive. For example, it costs several thousand MOP just to upgrade the IOS to support IP security. But why not use Linux on top of a standard x86 server as a router? To do that, we need to ensure the performance, support and ease of management. Now there is a company that aims to provide these.

Vyatta has put a trimmed down version of Linux on a standard IBM x86 server to create a high end router, at 1/5 of the cost of a similar Cisco router. In terms of permformance, it is rivaling a Cisco 7200 router. In terms of support, it provides all kind of support services up to 7x24 and consulting services. Regarding management, it has a command line interface that looks very similar to IOS, so you will feel right at home. For junior administrators, it has a web-based interface.

Vyatta has won the best of open source networking software award in 2009. As it is open source, you can always try it out yourself to see if it fits your needs.

Upcoming courses for network administrators

Any questions, ideas or experiences to share? Contact me at 88980601 or kent@cpttm.org.mo. We also have 3 more newsletters: CIO newsletter, Software developer newsletter and E-flow newsletter.

Until next time, 

Kent Tong