If you are one of these people raise your hand…

I’ll pause for a second while you ponder your response wondering why I just made that statement.

I’ll ask a few other questions now:

    Do you have a disaster recovery plan?

    Have you tested your disaster recovery plan?

    Have you tested a server recovery?

    Where are you going to perform DR?

    Who is going to be available to perform the DR?

    What level of expertise is going to be required to get things rolling again?

    What kind of hardware is going to be readily available for DR?

    Do you have a copy of all installed media at the vaulting company?

    Is there a written DR plan or is it all in someone’s head?

    Are you required to maintain data by law?

    How long can you be down with server X before it starts to impact company function?

    How long can server group Y be down before it impacts the company’s bottom line? 

Now are you ready for a Disaster?

My next question would be what do you consider a disaster (DISASTER)? Is it the CEO deleted a file and we need it back now, or is it the datacenter just flooded and ruined all my servers. Either way it is a disaster in someone’s life and you need to be prepared to handle it and know the amount of time and expense it is going to cost you in manpower and cold hard cash in these times of doing more with a lot less. 

I’ll ask another couple of questions:

    Have you ever stated “we never need to do restores” or “we have not needed to restore a file in 2 years, so our backups are not critical”?

    Have the number of servers and the amount of data being backed up exploded in the past couple of years?

    Have you been faced with the need to continuously buy more tapes?

    Do you need to expand your tape solution so that your backups fit within the backup window?

Many companies these days are cutting back on one of the most critical, but often the most ignored pieces of their data center. Backup solutions are never the most glamorous part the infrastructure and almost always delegated to the junior member of the staff to manage. However the backup solution is always expected to work when needed, no matter how little time or money has been spent to ensure its success. The backup solution is always the recipient of everyone’s wrath when it doesn’t provide the data that “should have been there”. Often it is ignored or will be addressed later due to other “more pressing” issues. 

What would be the outcome if half as much attention was paid to your DR solution as your email server, that 1 SQL database, or the newest application on the market that makes coffee and writes its own code, would you be able to sleep easier each night?

Backups are not sexy, nor are they the focus of most IT departments. There is typically a small budget set aside for purchasing tapes, but no major changes are usually planned. Most IT staffs do not want to be tasked with owning the backup solution so that the finger is not pointed at them when all fails. Most of the time backups fail due to very few reasons; configuration issues of the software are the most common. The second most common reason is the data grew too large for the backup solution to handle.

Over the past couple of years many new solutions have been introduced to be the magic pill for all backup issues. Some companies proclaim that disk based backups are the only way to go ignoring the need to get at least a backup copy of the data offsite. Many boutique backup solutions have been offered to handle specific applications, these solutions though typically do not integrate easily into most enterprise backup solutions. Electronic vaulting of data at remote location has been discussed and made a reality over the last few years as long as the pockets of the company are deep, purchasing the needed network bandwidth. I could go on and on with the various offerings that are available.

Having assisted many companies in the past 20+ years with their backup solutions, companies usually stop spending money and quit implementing at the “backups are running” point. Most intend to get the rest done when they can afford to spend the resources both physical and financial. Often Disaster Recovery looks like an unnecessary expense from the CFO’s perspective because many think that the backup is all that is needed for DR.

In these days of tight budgets, it’s a good idea to include additional funds for DR in new projects since it is going to be called upon to handle the additional load. Few companies include the price of DR into the cost of the newest application but place additional requirements on their backup solution to protect it. Many companies are implementing BI or data warehousing projects both of which are extremely taxing on backup solutions without including the cost of upgrading the backup solution to handle the additional load in the project.

For me the best solution is the one that I can count on to be there with the data I need to recover when I request it. There are no substitutes to good planning, solid implementation, great documentation and repeated testing. Did I say repeated testing? Yep!

So what’s it going to be — restful sleep each night or maintaining your resume?

• The drive cap per cluster is now raised to 250 drives
•  no cap exists when virtualizing IBM DS3/4/5 disk and ordering through Passport Advantage
• SVC EE FlashCopy is now included at no additional charge with SVC EE base virtualization
• SVC v5 with iSCSI support, performance improvements, and integrated solid state drives, among other enhancements

IBM Announcement on change of support withdrawal date.

I see the hunger to expand their understanding of the systems as a very valuable trait – a trait I personally wish was more dominant in our world. I praise administrators who actually take the time to read the CLI guide, discover that they can create server farms full of 4-way SMP VMs in their new VMware Cluster, and see the possible benefits of implementing IVR in their Cisco SAN environment. Knowing the abilities of the systems is a wonderful thing and can make someone a more treasured resource; however, it must be said that actually implementing something because you could, and not because you should, can be the source of some major headaches for yourself and others.

I’d be the first to tell you that most of my headaches are brought on because someone did something because they could – the system had the ability to do it, they researched how it could be done, and they pulled the trigger. Some of these features worked well for the first six months until they needed to expand the environment, and now they needed to take an outage to address the issue or pay for workaround to their implementation of that nifty new feature.

“ ’cause I could ” is possibly the scariest response to many questions I ask about why and how something has been implemented. Please, do us all a favor the next time you, or your buddy Chuck, plan to pull the trigger on some feature you just learned about—slow down and ask yourself a few questions first…Did you research how this feature impacts the other projects and systems currently in the works? Is this actually needed, and will it really do what you think it will? Did you pick the brain of someone with more knowledge about that particular technology and ask about the gotchas? A few more questions may just save a trip to Rite Aid for another bottle of Excedrin Migraine.

 Well you would be wrong. When you get the system up and running, you may notice that the drives are running in DEGRADED mode. This means that the system doesn’t see the aux cache card and to protect you from a SAS controller failure, it is kindly running without using any disk controller cache. Did I mention that it probably took you a LONG time to get to this point?

 Solution: Take a look at the server properties and you will notice that you have the SAS controller in T9 allocated but there is also a “SCSI Bus Controller” in C9 not being used. The C9 component is actually the aux cache card for the SAS Raid controller. Use Dynamic Logical Partitioning to add the C9 physical adapter to your partition and all will be well.

Jump ahead to the 1980’s and the PC revolution begins taking over, and in most cases, replaces much of the need for mainframe functioning due to the smaller computer’s ability to process applications efficiently at a fraction of the cost of the mainframe systems.  Since the computers “clock speed” or processor speed is low, there is very little heat compared to that of the mainframe.

In 1965, Gordon Moore of Intel created a fairly reliable prediction that was labeled “Moore’s Law”, which is, “the number of transistors on a chip will double about every two years”. Well that has happened fairly regularly up to the current time.

 
Problem with that is not only have the transistors doubled, but the die on which they are cast as well as the size of those transistors continues to shrink. We now talk in terms of “nanometers”. A nanometer (one billionth of a meter) is the measure by which technology companies use to define size of processor/transistor density.  To give you a point of reference to a nanometer, the width of a human hair is about 180 nanometers (nm).
 
Computer manufacturers are creating processors at the 65nm level now (32nm, by the way, is just around the corner). However, as we all know from our high school physics class, one of the major byproducts of electricity is heat (see paragraph one). And as manufacturers crank up the electrical speed of their devices, the heat becomes even more of an issue and increases exponentially, and has caused manufacturers to become increasingly focused on how to dissipate it.

Since we are dealing with transistors that are invisible to the naked eye, with components that are able to hold single eletrons, computer engineers have to start dealing with heat at a different level than most of us will ever have to understand – the quantum physics level.

At this level, heat envelopes and tolerances become a major factor since the material used to increase speed must be able to withstand chemical change (i.e., melting) for long periods of sustained use, as well as deal with the heat dissipation from components sitting next to each other.
 
As Moore’s Law continues, so must manufacturers continue to devise ways to cool the processers, and avoid us trying to devise “water chillers” at the nanometer level.

The thought that “If I increase my power, I should be able to drown out all that interference” is wrong and can be illegal.  It’s like saying “This party is getting a little loud, let me bring in my louder and more obnoxious friend Ted to make it better”.   I think that is likely to offend everyone else at the party and get you and Ted thrown out. 

One of the problems with amplification of RF signals is that the more you amplify the signal, the more you distort the signal.  So now, not only is Ted loud and obnoxious, you can barely understand what he says, his voice brays like a donkey and he stutters like a motor boat.  Now Ted’s annoying voice is making hard for some guys to hear what their girlfriends are saying and now they’re getting annoyed at you for bringing Ted in.

Now we are at the possibly illegal part.  Adding amplifiers and higher gain antennas are that were not specifically certified with the FCC by the device manufacturers is illegal.  Only the manufacturer has the right to certify anything with their systems.  No “professional installers” or anything else will make it legal.  Now you might get away with it since the toothless wonders known as the FCC do not do go actively looking for violations.  The FCC requires that information come from other wireless operators/users and that the wireless operators/users gather the spectrum analysis, photographic evidence, etc. that a violation exists.  So if you don’t annoy anyone you might be OK on this, unless someone from the FCC walks by and notices.  Once the FCC notices a violation you might be given some options like: make it compliant, take it down or have it confiscated plus others.  Also Remember, the FCC is part of the government and government never has enough money, so you might get a fine attached too.

But, you say, my 40,000dBi antenna was certified by the wireless manufacturer.  I would then refer you back a couple of paragraphs to “Ted”, and say “You aren’t doing yourself a favor with that antenna.   You would be better using this 2dBi antenna.”

With RF design is about science, balance and magic, not “More Power”.

To give you an idea of how fast a teraflop microchip operates, compare it to a simple desk calculator. It has an average speed of about 10 FLOPS. When we perform operations on these calculators, it appears the operations are instant. For a calculator to operate at 1 Teraflop, the calculation would have to perform at 10 to the 12th power.

To give a perspective of a trillion, a trillion seconds ago is 31,688 years.

Intel has introduced a prototype of a future processor with 80 cores capable, they say, of performing a trillion operations per second (1 Teraflop), at a power consumption of 62 watts at a clock frequency of 3.12GHz on a chip the size of your fingernail. Of course this is a future looking statement, so all bets are off until it is actually introduced at a reasonable cost to the market as a whole.

Apparently Moore’s Law still holds true – computing power will double every 18 months or so.