Discussing all things virtualization and storage in the data center.

2008 - one month gone already.

Its been pretty quite on the ole blog front and I put it down to the enormous workloads on people.  I think I have done more in the last month than I did during the last few months of 2007.  Unfortunately, that is not by design. 

One of my pet hates is back at it again and that is HP P Class blades and their enclosures.  Its amazing what a "hardware" fault in an enclosure can do to a SAN fabric and eventually storage on IBM DS4800's.  I found out that the DS4800 controller can reboot itself if it feels something is wrong somewhere and that can be a fault in a HP enclosure.  That means lots of systems lose their second path because of something that should not happen.  Wow, talk about a surprise to me and our monitoring team along with our senior management.  I have not seen anything like that on other systems. 

Its amazing how much time you can waste with troubleshooting in the SAN if you don't have any tools.  Especially when you decide to offer some of your knowledge to staff who just joined your team and have some basic idea of what a LUN but don't have a clue what a WWpN or a flogi might be.  I find it totally unbelieveable to be forced to take on staff who don't have any previous knowledge of storage and probably don't really care.  They were in the backup (CommVault) team and now belong to my team.  This is where backups need to be isolated from the real work of storage management.  I think too many people think writing to a tape is akin to disks....  But then again, management decisions or better still indecisions are always mysteries to me.

This leads me to HCAP.  We have it but no one wants to know a bar about it.  My management keeps wanting to get it running in a dev environment but what good is that?  I can't very well go changing our document management or messaging interfaces to say "archive this data as it is important to the long term security of our organisation".  None of the people in charge of document management or messaging are the slightest bit interested in HCAP as they "ARCHIVE" to tape....  I desperately hope that some really really really important file or email goes missing just before some litigation happens in 8 months time.  That will definitely change the views on archiving. 

Electronic document management is still not as good as the old registry with paper files.  They were the good ole days and even now, registry is still responsible for their own archival especially in government departments.   A lot of scanning still goes on but I often wonder where the soft copies are kept.  Lets hope it is not on a SAN drive that gets backed up to tape...

I am having a meeting with HDS's HCAP marketing guru on Monday to discuss exactly what HCAP is good for as far as HDS goes.  Everyone tells me HCAP is built on openness.  That means the end user has to do something with it.  I am sure everyone has the time it create systems built on openness.  I am already converted.  Pity about the other thousands of people in our organisation.

One last thing,  You can't use RAID 6 when you are using concatentated parity groups in the USP/USP V.  Thats a sad state of affairs if you ask me.  I feel comfortable using larger disks and concat parity groups to gain performance but larger disks create stress after failures when reconstruction occurs.  Can you imagine a double disk failure in a 4 way concatenated parity group that is using 300 GB plus disks?  To further compound the issue, LUSE could also be involved and we can be restoring many many TB's of data.

Stephen

comments powered by Disqus