What happened at IBM Highbrook datacentre?

Forums › IT Pro and developers › What happened at IBM Highbrook datacentre?

Mark

1360 posts

Uber Geek
+1 received by user: 345

Topic # 114451 20-Feb-2013 11:50

Anyone in the know as to what caused IBMs outage yesterday ?

[Moderator edit (MF): Moved to correct forum, changed subject to something meaningful]

1 | 2

Awesome
4810 posts

Uber Geek
+1 received by user: 1062

Trusted

Subscriber

Reply # 766332 20-Feb-2013 11:53

Is it fixed yet? I haven't seen an updated article.

I'd like to know this myself.

Twitter: ajobbins

graemeh

2078 posts

Uber Geek
+1 received by user: 230

Subscriber

Reply # 766334 20-Feb-2013 11:56

According to ComputerWorld it looks like even IBM don't know :)

PIERCD

58 posts

Master Geek

Reply # 766341 20-Feb-2013 12:09

Watch out for the Big Blue Curtain of Silence.

Took a while for the reason behind the Newton Data Centre falling over to come out as well.

ajobbins

Awesome
4810 posts

Uber Geek
+1 received by user: 1062

Trusted

Subscriber

Reply # 766345 20-Feb-2013 12:15

PIERCD: Took a while for the reason behind the Newton Data Centre falling over to come out as well.

The full story about these kind of things never makes the public domain. I'm quite sure there is a good reason Air NZ shut up very quickly after Fyfe's initial outburst.

Twitter: ajobbins

PIERCD

58 posts

Master Geek

Reply # 766356 20-Feb-2013 12:23

ajobbins:
PIERCD: Took a while for the reason behind the Newton Data Centre falling over to come out as well.

The full story about these kind of things never makes the public domain. I'm quite sure there is a good reason Air NZ shut up very quickly after Fyfe's initial outburst.

Agreed, but the industry is so small, it won't take long for it to leak amongst those who know people.

Of course whether that information gets posted to a public forum is a different story. I know my employer would not be happy if I posted about it here (and we didn't even supply any product to that Datacentre).

Mark

1360 posts

Uber Geek
+1 received by user: 345

Reply # 766378 20-Feb-2013 12:57

PIERCD:

Of course whether that information gets posted to a public forum is a different story. I know my employer would not be happy if I posted about it here (and we didn't even supply any product to that Datacentre).

Create a disposable account and post it up ;-)

ajobbins

Awesome
4810 posts

Uber Geek
+1 received by user: 1062

Trusted

Subscriber

Reply # 766395 20-Feb-2013 13:30

Resolved:

http://computerworld.co.nz/news.nsf/news/ibms-auckland-datacentre-issues-resolved

Twitter: ajobbins

Mark

1360 posts

Uber Geek
+1 received by user: 345

Reply # 766432 20-Feb-2013 13:55

Can't wait to see what the root cause turns out to be :-)

gzt

10193 posts

Uber Geek
+1 received by user: 1562

Reply # 766469 20-Feb-2013 14:52

Reading the press my feeling is multiple failures not just one cause.

I will hazard a blind guess that logistics also played a part in the long delay.

insane

2280 posts

Uber Geek
+1 received by user: 370

Trusted

Subscriber

Reply # 768422 24-Feb-2013 10:29

Given it was just the virtual platform which fell over and not the whole DC, and that it happened at 3am I'd imagine a storage firmware update or failure. From experience if power is not an issue, then only a storage issues will take you completely out like that.

Reading some of the comments by others in the media about why IBM didn't have another DC etc is hilarious. If customers were wiling to pay they would have bought a DR service and been covered, but they didn't, they bought services from a single site and guess what, sometimes stuff goes wrong.

*EDIT*

Someone over at computer world wrote this:

Fact #1. It wasn't XIV storage.
Fact #2. It was not related to planned work or changes - it was a fault.
Fact #3. No data was lost.
Fact #4. There was some corruption with a small subset of data which was successfully restored from recent backup.

gjm

747 posts

Ultimate Geek
+1 received by user: 91

Reply # 768426 24-Feb-2013 10:43

insane makes some reasonable assumptions ...

[Amstrad CPC 6128: 128k Memory: 3 inch floppy drive: Colour Screen]

Zeon

3409 posts

Uber Geek
+1 received by user: 404

Trusted

Reply # 768436 24-Feb-2013 11:10

insane: Given it was just the virtual platform which fell over and not the whole DC, and that it happened at 3am I'd imagine a storage firmware update or failure. From experience if power is not an issue, then only a storage issues will take you completely out like that.

Reading some of the comments by others in the media about why IBM didn't have another DC etc is hilarious. If customers were wiling to pay they would have bought a DR service and been covered, but they didn't, they bought services from a single site and guess what, sometimes stuff goes wrong.

*EDIT*

Someone over at computer world wrote this:

Fact #1. It wasn't XIV storage.
Fact #2. It was not related to planned work or changes - it was a fault.
Fact #3. No data was lost.
Fact #4. There was some corruption with a small subset of data which was successfully restored from recent backup.

Agree completely. People should have always put in a complete seperate DR.

khull

1245 posts

Uber Geek
+1 received by user: 137

Reply # 768463 24-Feb-2013 12:33

You'd be surprised how many clients have little regard for DR, or clients who only want to pay for a SLA and then subsequently add loading beyond what they are willing to pay for

exportgoldman

1200 posts

Uber Geek
+1 received by user: 3

Trusted

Reply # 768492 24-Feb-2013 13:38

This was posted over in the comments at Computer World.

Ok, So here are the real facts....
Their IBM XIV storage was due an upgrade, a new cash card was installed that was faulty and it took the entire array down. It then took 3 Days to re-store the array from back-ups. This has come from a customer effected by the outage.

So much for XIV being Tier 1 Storage.... In my opinion, there are only 3 suppliers that produce Enterprise Grade Storage in the IT market: HDS - VSP; EMC - VMAX & IBM - DS800 range... Everything else is midrange with single points of failure.

If this is indeed true, and it fits from my experience (only Storage can take everything offline for nearly 3 days, with complex VLAN Networking problems second.) Then it looks like this was not a well engineered solution, or alternatively hit bugs in the XIV Storage Array which they needed to get programmers to resolve.

If the comment from Computer World is true, then I would have expected IBM to have a second XIV Storage Array which they would have been doing snapshots to, and could have failed over to, before they did the hardware upgrade on the primary XIV Storage array, then after testing they could have fallen back to the primary and upgraded the secondary array.

It appears IBM only run 1 storage XIV SAN Array in that case.

I know of clients in the graphics industry which buy two of these for exactly these reasons, so if IBM didn't have a second one you wonder how cheap they really are.

This is of course all speculation - as I have no insider information, just experience working on a range of SAN's.

Tyler - Parnell Geek - iPhone 3G - Lenovo X301 - Kaseya - Great Western Steak House, these are some of my favourite things.

gzt

10193 posts

Uber Geek
+1 received by user: 1562

Reply # 768502 24-Feb-2013 14:06

Remember that IBM indicated two faults. One which caused the original issue and another which was discovered or which occurred when they planned to resume service. It is possible there are several sets of correct information floating around relating to each fault. Your quote and insane's quote may be consistent. Both were anonymous.

One thing is for sure, the continuing speculation is not healthy for IBM reputation.

A conspiracy theorist might suspect this calculation has been made already and details of the fault and the issue found in resuming service may be more damaging than continuing speculation. ; ).

The other perspective is this issue is hardly anyone's business and we should just do the proverbial.

1 | 2