Storagebod Rotating Header Image

Web/Tech

Expo Reprised

Oct 17th, 2008
by Martin Glassborow.
No comments yet

All the vendors I chatted to were fairly upbeat; not as busy as last year but most of them thought there were better quality leads than in previous years. 

1) Pillar's stand seemed to be smaller than last year.

2) 3Par's t-shirts seemed to be a little faded.

3) There's been a fair bit of movement between vendors as per usual.

4) I didn't even get to the Copan stand, must have missed it.

5) No one wanted to talk to me on the HP stand, don't think they like me!!!

6) There were a few bleary eyes and imagine more this morning!

7) No-one tried to sell me Microsoft's Storage Server, so no-one got laughed at!

Disappointed to miss Barry Whyte on IBM's stand. IBM are really pushing XIV; got a few more roadmap details  about the next release. I'm still not convinced and asked the question about double failure; didn't get an especially satisfactory answer apart from it being statistically unlikely.

I think the DS8k range is a better product, I really do! I understand that IBM just haven't got the penetration into the Open Systems market place with it, will XIV do it for them? We'll have to wait and see; there certainly seems a willingness to start the sales-pitch with no trousers on but I don't think price was ever really an inhibitor for the DS8k.

Lots of people talking about FCOE, there's a real buzz about it.

And, oh yes….I actually won something. I won a Flip video-camera on the EMC stand, I've had one in my shopping basket on Amazon for yonks but not bought one and now I don't have to!

And rather more people are reading my blog than I realised! Thanks everybody, it's nice to know that people do read it and more importantly, people enjoy it and I at least bring a smile to some faces!

Why no Ontap CLI (client)?

Oct 9th, 2008
by Martin Glassborow.
1 comment

Now I have posted about how easy a NetApp array is to configure and manage via the web front-end and even the CLI is really nice and consistent but I also have a problem with the CLI! There is no client CLI, you have to log onto the array to run the commands; this makes scripting a real pain, even more painful when ‘rsh’ protocols are not allowed in our environment (and rightly so).

EMC, IBM, HDS et all all provide a client-based CLI; why can’t NetApp do something similar? Can I make a plea to my NetApp readers; please provide us with an equivalent to the CLI environments that your competitors provide; without this doing job control using things like Control-M is going to be a bit of a pain.

It kind of surprises me that there isn’t one already tho’; I would have thought most Enterprise customers would want one.

Dispersed Storage

Oct 6th, 2008
by Martin Glassborow.
4 comments

Every now and then, I actually read some of the ‘spam’ which gets sent to me; this one had a link to a company called Cleversafe; I thought it sounded interesting, so I thought I’d click and have a look.

What does it do? as far as I can tell, it takes your data and slices it up; uses an information dispersal algorithm to distribute these slices either locally or to another remote site.

The theory is that you can define the number of slices and how many slices you would need to reconstitute the data. So I could distribute my data across a number of data-centres but only need a sub-set of those data centres to access my data.

I could have a number of data centres distributed in seperate countries, none of which hold enough slices to reconstitute useful data. This would protect me against a government seizing my data centres and getting any useful data.

Yet again, I think we have a solution looking for a problem and some of the stuff on the website is just bizarre. It compares itself to RAID-3? Who the hell uses RAID-3? I know it actually looks very similar to RAID-4 but by comparing to RAID-3 you kind of have a bit of credibility issue and it makes me feel very edgey about dealing with you.

And I am intrigued with their claim that with enough low-latency bandwidth that their storage can be as fast as local storage. Obviously this is entirely possible but surely you have to be within the normal synchronous limits for this to be true and this kind of obviates some of their claims to protect against corrupt governments breaking into your data centre and getting access to your data.

I am interested in dispersed/distributed data storage as at the moment, I am wondering how to store multiple petabytes of digital data and how to protect it; this looks interesting but I do wonder about the company!

Chuck, does MAUI do something similar? I missed the presentation EMC did for us last year and I’ve not got round to arranging another. Anyone else reading got anything worth looking at?

How to annoy a customer….

Oct 3rd, 2008
by Martin Glassborow.
2 comments

On occasion, vendors do something which really want you to take them outside and shoot them.

I have a case where a certain vendor’s product which is key to some of my applications caused me some real problems about this time last year. A work-around was produced and we are happily working again. Now we have to increase capacity in some key environments and we need this product to work against the latest and greatest; so in goes the RPQ and back comes the answer; ‘Sorry, product is going to be withdrawn but product ‘Y’ from vendor ‘X’ will do what you need!’ 

Okay, so you pointed out an alternative but it’s going to cause a huge amount of work; a shed-load of political fallout and general pain all round! It should at least have been accompanied with an offer to work with us to transition to the new product, gratis of course!!!

Rant over!!!

Light blue-touch paper and stand well back

Sep 30th, 2008
by Martin Glassborow.
10 comments

A friendly NetApp bod sent me the link initially but I see that our friends at The Register have already picked this up, NetApp’s 50 per cent guarantee.  I think I’ll enjoy the fireworks!! I can see a number of points of contention

  • Comparing against RAID-10; this is going to cause some pretty explosions
  • Only 10% of the data can be things which can be hard to dedupe; perhaps the odd rocket here and there
  • Its for VMware environments only (article misses this I think); a nice jumping jack
  • Utilisation of Snapshots; twinkly catherine wheel
  • No high performance workloads which require lots of spindles; whoosh, look at the Roman Candle
  • Must use NetApp PS engagement; oh blimey, is that the burning effigy of a NetApp spokesman or just some Guy? (For my non UK readers, we traditionally burn an effigy of Guy Fawkes on a bonfire night)

Now, I’m not making any judgements but flipping heck….this is going to be colourful!

But when did storage get so hard?

Sep 30th, 2008
by Martin Glassborow.
1 comment

In contradiction to my previous blog, storage has now got a whole lot harder! In fact the sheer ease of implementation is making the whole process a lot harder. We can allocate/implement storage incredibly quickly these days; this is leading users to leave allocation till the last minute (they always did anyway) but now we can actually do the implementation really quickly but things are going in without proper thought and design.

My team’s OLA is two days from request to having storage presented; we can do it quicker but we try and ensure that change management is followed. So that bit has got easier but now we’ve got all kinds of other problems to deal with.

  1. Which tier of disk, before the user would have a choice of Tier 1 or Tier 1.
  2. Yes, I know SATA disk is much cheaper than fibre and you can get them for £120 in PC World next door, please feel free to put your business critical database on them but good luck with that.
  3. Snapshots vs Clones and no they do not replace back-ups.
  4. Replication; easy to implement but please understand that this does not replace your offsite back-ups by itself. And no, we can’t give you a zero data-loss option if you want asynchronous; well not without some kind of performance impact.
  5. Performance of the arrays is much better, so people just throw things at the array and hope that it will perform. We used to spend hours, days and weeks laying out arrays to get the optimum performance; now we don’t get time and then when it all breaks, it’s a big mess to sort out.
  6. Agile development methodologies which seem to be concentrating just on functionality and not NFRs. I need to know how many IOPs, read/write ratios, random/sequential to enable me to make a better decision as to where the data should live.

Some of the above is just down to educating users but some needs the vendors to work on and give us better tools to allow us to work faster and make better decisions.

You’ve made the grunt-work easier, congratulations. You need to make the really hard stuff easier as well.

We need better tools to monitor performance and proactively fix problems before they become problems. Good information about the performance of products; meaningful performance figures from a vendor with regards to speeds and feeds (SPC-1 gives me a high level indication as to how fast your array is for that known workload, unfortunately SPC-1 is not actually a business application) or at least a tool which allows me to model a workload easily to work out quickly how many spindles I need, how much cache I need etc. BTW, I know the tools exist; your pre-sales support guys have them, I need them too.

This is commercially sensitive information I hear you cry! So what!? It is the information that I need to do my job and maximise the return on investment. I don’t want to have to bother you when I need to do this bit of modelling; I certainly don’t want to pay you whenever I need to think about putting a new workload on an array. And yes, I’d really like my answer by Close of Business today; yes, I do realise that it is now 16:00.

So can we now have some tools which make the planning and design of storage infrastructures easier? For free would be nice as well!

End-users and the things we do!

Sep 24th, 2008
by Martin Glassborow.
8 comments

‘Zilla is so right with this post; it makes me laugh, it’s exactly how we think. And I think EMC miss a big trick at their customer council events, they should ship a couple of TC/SEs to them; tell them to sit in a corner and just listen to the conversations which happen after about midnight and we think there are no EMCers around. It’s funny, there is always EMCs official theme and then there is always an underlying customer theme and probably not always something EMC would approve of.

Last year for instance, there was a lot of discussion about putting NetApp heads in front of EMC disk. Not supported by EMC for sure but there’s enough people out there who have tried it with success, that it doesn’t scare us. Us end-users do all kind of things which aren’t supported; we do try and avoid being first but there’s always someone who is rash enough to try it.

p.s it’s not just EMC customer events, I’ve sat on the IBM customer council and exactly same thing happened.

Who Pays for the Cloud?

Sep 22nd, 2008
by Martin Glassborow.
7 comments

This is a good post on the above question; it nails one of the biggest problems that we have internally; building out our internal cloud so that we can provision from it. Not technically an issue, we know how to do this, well more or less!

Our projects tend to be point-funded; charge-back to projects is also hard to implement; this is not a technical problem, this is a budgetary problem. Building cloud infrastructures does not fit with with either current budgtary practises but also it does fit well with the current annualised budgetary cycle.

What Keeps Me Up At Night??

Sep 22nd, 2008
by Martin Glassborow.
No comments yet

One of the many questions I get asked in one form or another is ‘What Keeps Me Up At Night?’ This is a great question for a vendor to ask and I will pontificate for hours on what is wrong with the storage industry and how I’d fix it! The question can be split into two general themes, ‘Technical’ and ‘non-Technical’. And under ‘Technical’ comes Better, Faster and under ‘non-Technical’ comes cheaper. And often the latter overides the former; a product can be Better and Faster but if I can’t show a pay-back in less than eighteen months and twelve months ideally, I am going to struggle to get any kind of funding.

Over the next few months or so, I am going to examine things which could allow me to do things Better and Faster; some might even allow me to do them Cheaper.

Newer Posts →