Storagebod Rotating Header Image

Dear Santa, can I have a good Management Tool?

Well, it appears that I've hit a bit of nerve, my post on ECC was my most popular post ever. Now, it probably didn't come out as clearly as it should have done; nearly without exception

'All your products are rubbish!' 

There you go, I've said it now! ECC gets picked on as it is probably the most deployed piece of rubbish. It's also a horrendously expensive piece of rubbish and I do like Ian's comment even  "if ECC was truely free, would I still use/deploy it?". That should worry EMC because I know who Ian works for and they are a truly huge storage customer but if they won't listen to them, the rest of us stand little chance.

But some hints and tips, focus on the management aspects of your tools. Most of you have pretty good administration/provisioning tools and a great number of us simply use your CLIs to provision anyway. Yes, a common GUI for provisioning would be nice but it's not necessary. Instead concentrate on getting  

  • Alert/Event Management Working
  • Capacity Reporting
  • Performance Reporting
  • Configuration Tracking
  • Scalable, it needs to manage upto and this is an arbitary figure, 5 petabytes and it should take account that a lot of us still have lots of small LUNs.

Auto-discovery of storage devices would be nice but it's not necessary for most of us. Ian, with his several gazillion arrays might be different.

Have a good look at Onaro, it's not perfect but it's probably the closest to being a good Storage Management tool out there.

And storage array vendors, I don't expect you to give away your IP but documented APIs which give read access to information from the arrays would be good. Random changes to CLIs are bad, they break our scripts and they break tools who use the CLI for their information gathering; so play a little fair.

Anyway my thoughts, I'm sure you have your own ideas; please chip in.

        

 


10 Comments

  1. Ronald says:

    I betcha a pint of Guinness that the vendors won’t respond on their blogs about their rubbish management.

  2. marc farley says:

    Ronald, is that serious bait? Because some of us vendor types would go to great lengths for a pint of Guinness.
    3PAR doesn’t have global management software, but we do have a relatively inexpensive management tool called System Reporter (I think the price is $5K US) that provides a lot of historical and real-time performance data. It’s a metering application that customers use to get at I/O data for all array components, including individual drives, virtual volumes and all I/O ports. Gives you service times for all volumes. Its a diagnostic tool for understanding how the system is behaving, but we sell it to to customers so they can see the exact same information our service organization and SEs see when they look into the guts of an array. It’s not the end all for storage management, but its an incredibly valuable SRM tool for 3PAR systems that most people are pleasantly surprised to find out about.
    The thing I’m not so sure on right now is how it is used. There is a CLI and a GUI, but I don’t think there is an API.
    I’ll try to get more info on this and put it on my blog tomorrow after I wake up. Its late here now and I’m going to bed.

  3. Martin G says:

    Systems Reporter looks nice, just needs to be bigger and more chocolately. And obviously needs to support more vendors.

  4. inch says:

    heh..
    Is it possible to grab your details off “list”? 🙂

  5. Barry Whyte says:

    I’m not going to defend IBM’s SRM tool… but its difficult to do all the things you ask without installing agents on the hosts… but ’nuff said.
    As for CLI – absolutely. Our SVC CLI ethos is to *NEVER* change the syntax, only add to it, but if something isn’t specified to accept the default and never break scripts… believe me it kills the test department first, I’m using some 7 year old scripts that still do what it says on the tin.

  6. Martin G says:

    I don’t object to installing agents. It is better than some of the alternatives I’ve seen!
    And I know it’s hard but hard things are worth doing!

  7. Ian says:

    AGREE. (except would it be more realistic to ask Santa for “a working mngt tool” – he’s only a mythical figure let’s not ask for miracles..)
    Can I be greedy and add a couple of things to the list :-
    * Multi geography
    * Estate analytics
    * Cover storage devices (not just disk arrays)
    These are not direct mngt reqs but are absolute reporting requirements for large companies.
    But the overall principle should be – get the ground level basic stuff (is SRM) working before dreaming about other buzz word / hype mngt.
    It’s increasingly hard to justify the cost (the full FTE, infra, licenses, impact etc) of ‘mngt’ technologies Vs FTEs and CLI/scripts and home-grown stuff – trust me I know which I’d prefer but the justification is increasingly unrealistic 🙁
    Re Agents – they are fine PROVIDED, we don’t have to buy more CPU/RAM to run them, they are security aware and don’t need 20 IP ports opening, they can be self upgrading/managing and they don’t impact the interop stack requirements.
    I accept that nobody has a product today that either works, meets all the requirements or is cost effective – but what worries me more is that I don’t see (even under deep NDA) anything in the pipeline that addresses this…
    Perhaps (as requested to a number of partners) the use case, volume and feature set reqs for SRM products could be validated with customers prior to the dev droids starting work?
    As Martin says, have a deep look at Onaro – ask Onaro as to why their product works so well, solves real problems, is rapid to market and adopted in so many customers – it’s because they engage & listen to the customer (not preach to them), it’s a culture thing.
    And don’t get me started on being asked to invest (mindset, $s, function, interop etc) for other required technologies as add-ons where the real cost is exposed once you’re committed to the partner (eg backup reporting (just been told that’s a 7 digit cost in 09), multi-pathing…)
    “I don’t want any more storage mngt crack – time for cold-turkey and rehab me thinks”

  8. Mike Shea says:

    Nice post – I was a Solution Architect for ControlCenter (ECC is an IBM trademarked thing – and EMC does not use it). It is a big, bloated, piece of SW with a ton of moving parts. And I mean a ton, more than any customer is ever told of.
    What always gave me the willies was having to have that first discovery meeting where I went in and played twenty questions in order to do a quick whiteboard proposal on an architecture. Of course at this time, we discussed the requirements of new code on their switches, HBA’s, Symms, CLARiiON, Navisphere, Possible OS patches and application patches if they desird to roll out application agents – which by the way I found to be very useful.
    Of course all this requied scheduled downtime and middle of the night work by the now hopelessly inflamed admin team. It was always brutal visiting them again. It truly sucked to be me. 🙁
    Installing it was three projects
    – first collecting detailed information on the environment – things like code levels, patches, etc etc on all hosts, HBA’s, switches, arrays etc. All this gave me the ability to architect your solution and fully document what you needed to do in project 2 and 3.
    – second, upgrade all the code levels across HBA’a, switches, arrays, etc.
    – third, finally install or upgrade ControlCenter.
    ControlCenter does a lot of things, There are a ton of features in it and 90 ways to do the same action all the way through it. The learning curve was pretty steep, and it always amazed me that some would climb it pretty effectively.
    Truth be told, management is a real blind spot in the entire storage industry.
    The Holy Grail is the desire to manage a mixed environment. Onaro (Now called SANScreen) comes real close to getting it all under its belt, and I know they are laser focused on being complete.
    I like the product a lot, but hey, I am biased – I work for NetApp. Grabbing SANScreen was a very smart move.
    Cheers all – nice blog!

  9. Don’t want to give you a sales pitch here, but you might want to take a look at our monitoring and reporting tool, Tek-Tools Profiler (disclosure – I’m a product manager for Tek-Tools).
    Profiler addresses the issues you’ve bulleted above. Our goal has been to fill the gaps left by native tools and give users the broadest visibility into their infrastructures possible.
    We’ve got some demo videos, information, and free evaluations at our web site: http://www.tek-tools.com. If you’re interested, take a look.
    I’d be interested in your feedback, particularly the challenges you face day-to-day in managing your storage.
    Happy New Year! Great blog!

  10. i really like this post… thanks!

Leave a Reply

Your email address will not be published. Required fields are marked *