catsidhe | Oh, FFS.

The University's Grand United Computer System — on which is done trivial paperwork like receipting, leave requests, training applications, publications recording, that sort of inconsequential nonsense — is shortly to receive an upgrade. From very soon, those who need to do super-sekrit special financial stuff will no longer need to have JInitiator installed on their machines, but will require Sun's JRE.

And there was much rejoicing, not least because this is already in many departments' SOEs, and we thought that this would be a cakewalk: for most people the job was done before they started.

Then the instructions arrive:

uninstall any existing installation of the JRE.
install this particular, obsolete, version of the JRE (and we're talking more than a major revision old, here)
turn off the JRE autoupdate function

*headdesk*headdesk*headdesk*

Luckily, everyone I personally work with also saw the minor flaw in this plan immediately.

But still, WTF were the twerps in Central IT development thinking??!??!!?
On second thoughts, don't answer that.

Flat | Top-Level Comments Only

Despite that, the upgrade will be forced through, pain and suffering aside.

That is pretty much what you forsee, is it not?

Oh, it's a foregone conclusion. We know that it's a step backwards, but that doesn't mean that we're allowed to not do it.

I just feel very weary.

This surprises you for what reason, exactly?

Next up in our Adventures of the Stoopid: trying to persuade a certain company to do weekly full backups of their databases, with incrementals (or differentials) in between. Because if they try to shovel another 20 TB of data into the backup system each night, as they're planning on doing, it is Not Going To Work.

On the plus side, this happens two weeks before my contract expires.

On the minus side, this happens two weeks before my contract expires.

Who says I'm in any way surprised?

Just because I'm appalled, doesn't mean I'm surprised.

Searching for alternate unRecovery is starting to look really good at times.

Maybe you could do some envelope-back calculations and show the point at which each nightly takes longer than 24 hours to run...

It's not the nightly. It's the processing TSM does during the day that's the issue at the moment. It starts at 5am, and I've known it to keep going until 4pm. Which leaves very little time for things like reclaiming tapes and similar.

I'm starting to think that a beefy box (Sun, Linux, whatever), with lots of 8 lane PCIe slots, and lots of quad port 4 Gbps HBAs, and lots of LTO4 drives, is the way to go. Which will run into the $hundreds of thousands, I have no doubt.

One LTO4 drive == one 4 Gbps port. Then you need the disk to spool the data at native speed. The current workload, based upon the current drives' native speeds, could be handled by 8 LTO4 drives, but given that so many databases backup direct to tape, and I'm not convinced they can stream LTO4 off their disk, I'm thinking 12-16 would be better. Per site. Two sites. How much would you expect to pay for 24-32 LTO4 tape drives?

You don't by chance work in the said CENTRAL IT unit do u??

I recall this exact issue with the old learning management system. As I recall it was one of the justifications for not allowing any new subjects onto the system, for fear of backups needing more than 24 hours to complete.

Not that it matterred for long with the new learning management system taking over from it... which is why they didn't invest to resolve the issue.

-- mpp

Nope. I work for a small computer firm, whom you've probably never heard of. They've had some modest success manufacturing CPUs for game consoles in recent years.

The fundamental issue with this particular client is the ridiculous amount of data being backed up. Moving to incrementals on a daily basis, with weekly (or even monthly) full backups would go a very long way towards resolving this.

The situation is not helped by having (to give one example) three Sun T10000 drives (same throughput - on paper - as LTO4) hanging off a single 2 Gbps port on the backup server.

The aforementioned beefy box is basically throwing hardware at the problem. Not my preferred solution - I want to see a serious look taken at the systems to find out just why the hell they're storing so much data - but anything outside that is beyond the scope of my job (and frankly, I really don't want to have to deal with anything outside that - my job involves enough tentacles as it is, thank you very much!)

Coincidentally, a key part of my Univerity's Grand United Computer System is going down for an upgrade shortly, taking with it simple things like leave, salaries, and other trivial matters. I'll be interested to see if it comes back up demanding obsolete versions with which to work.

It's a government conspiracy to reduce the cost of education by ensuring no evidence of your entitlements exists.

-- mpp

...nope. I'm not. If Dilbert were not documentary rather than satire, why would Scott Adams be a multimillionaire????

Besides. I studied at Uni for 12 years and worked there for 9. And NOTHING they do would surprise me any more......

Oh, FFS.

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

no subject

I wish I could say I was surprised but......