shopify
analytics tool

Defining an up-to-date TCO Model

I am currently writing a TCO piece for MatrixStore against tape, optical and other disk based solutions. TCO studies are rarely independent and naturally favour the client who paid for the study or the organisation that the person writing the report works for (like me … Tape booo, MatrixStore yeaah).

We at Object Matrix would welcome an open forum on calculating TCO with other Vendors but realistically that is not going to happen. Martin also rightly pointed out that defining what should be included in a TCO model will be a challenge so I have decided to bullet point the core elements of a TCO model that I will use and I would ask others to contribute…

TCO Model Traditional Elements

  • The Analysis period (should include time to ingest and exit not just time content sat on media @ianhf)
  • Initial storage requirements
  • Annual growth requirements
  • Base hardware and software costs (RRP or Discounted @timothyburton)
  • Capacity % utilised of storage media
  • Capacity upgrade Unit Size and frequency (@ianhf)
  • Annual predicted % change in cost of media
  • Media Rotation costs (tape and spindles) @thesharmanator
  • Verification of written data costs / proof of ongoing authenticity
  • Replication cost (comms link vs manual actions)
  • Networking costs/Bandwidth utilisation (LAN/SAN etc). Network hog for minimal gain? (@ianhf)
  • Employees required / costs / % annual change
  • License costs/transferable? (@ianhf)
  • Training/Certification costs of operational/support staff (from @Hugh_Waters)
  • Energy usage / energy cost rises / better future energy efficiency % changes
  • Servicing costs / employee training
  • Failover facilities
  • Floorspace (from @storagebod)

Non -Traditional Elements that must be included

  • Annual value to company of assets being available online | Efficiency | Revenues
  • Annual depreciation or gain of asset value (@ianhf)
  • Energy cost of staff managing the solution
  • Estimate lost revenues following data loss
  • Cost of downtime (staff etc) following data loss (from Mark Wood)
  • Exit Costs (moving away from the solution and secure disposal) (from @storagebod @ianhf)
  • Maintenance costs | changes to firmware/drivers of connected systems (from @storagebod)
  • Cost of disruption of firmware / software upgrades (via @ianhf)
  • Costs of verifying end-to-end supportability (from @storagebod)
  • Application integration/certification costs (from @storagebod)
  • Cost of ongoing infrastructure interoperability validation between 3rd parties (via @ianhf)
  • Cost implications of Re-purpose existing tin or buying 2nd hand (from @skirrid)
  • Cost of restoring content (from @timothyburton)
  • Cost of disaster recovery options (from @timothyburton)
  • Flexibility of license model per site, per node or per asset (@ianhf)
A while back we plugged most of this info into a spreadsheet which we are updating also. Its an open sheet for people to put in their own values. When ready we will publish and ask for feedback (good or bad).
The View from OM
The truth is out there…
2018-05-07T14:01:39+00:00 September 17th, 2009|

19 Comments

  1. Martin G September 17, 2009 at 2:08 am

    Looks pretty good.

    Howabout exit costs? Including costs of disposal?

    Floorspace?

    Total maintenance costs including things like changes to firmware/drivers of connected systems? Costs of verification of end-to-end supportability. Application integration/certification costs?

  2. NPT September 17, 2009 at 2:15 am

    Cheers Martin, have updated.

  3. Hugh September 17, 2009 at 3:55 am

    Might training of operational/support staff be a consideration?
    If new storage method needs only 5 mins to learn how to manage, it costs less than one that takes a week long course to learn (which may also need repeating when staff change of course).

  4. Mark September 17, 2009 at 7:24 am

    Cost of staff down time whilst data is being recovered if a data loss occurs.

  5. Tim Burton September 17, 2009 at 7:52 am

    I am very much of the option disk > tape, however am a Tivoli SM admin and feel tape isn’t given a fair fight, so I’ll be in the devil’s advocate red corner. 🙂

    Power and cooling, very important for soho c(l)ustomers.

    Restore times (+ prioritised data)

    Hardware redundancy / DR Site integration

  6. Glenn Hall September 18, 2009 at 5:45 am

    Nick, the top argument isn’t right. Its not TCO for OM versus, tape/disc/nearline &c. Its more (imho) how OM fits across a real deployment situation. Most folk, have some of most things. Good though the notion may be, they aren’t all that likely to go all the way in one direction.
    So Part of the overall TCO could be to create a new factor to measure positionality * in an overall deployment. Is it hard to deploy.. in _your situation_? Can you migrate from and to in _your situation_?
    * means a factor to indicate relevance of OM in the whole solution. Scale could be 30(hi) where 0-10 is hard, 11-20 is time to start and 21-30 is delivering value.
    @skirrid

  7. NPT September 18, 2009 at 7:22 am

    Thanks Glenn. A TCO model needs to be all things to all people really.
    Aim of this post is to first get the elements right and then
    Allow people to prioritise the elements most relevanr to them.
    Vendor stuff can be plugged in then to produce a relevance score..

    This aint going to be solved soon but a goo starting point.

  8. ianhf September 21, 2009 at 10:44 am

    Some additions re TCO :-
    * cost of ongoing infra interop validation between 3rd parties,
    * frequency & disruption of firmware / software upgrades
    * size of units or growth (modularity)
    * disruption of growth units
    * performance measure (eg TCO per IOP per GB etc)
    * for capacity the difference between marketing GB and accessible GB (after b2->10, system overheads, resiliency etc)
    * flexibility of functionality license models (eg site or per asset)
    * if functionality licences transferrable during refresh
    * install & deploy costs
    * ongoing relationship mngt costs with vendor
    * cost of secure erase (either media or controller assets)
    * cost of vendor remote mngt solution (inc auth, logging & comms)
    * cost of mngt interface ports in network

  9. ianhf September 21, 2009 at 10:54 am

    and another :-

    * need to understand the expected useful life of the asset (inc waterfall tiering)

    * the TCO model must be over that period +1yr before start (to cover ingestion) and +1yr end (to cover egress)

    * the 3rs party support and maint must also factor in any ‘doubling up’ costs during ingress/egress (ie refresh)

  10. ianhf September 21, 2009 at 11:40 am

    and yet another one :-

    * bandwidth utilisation (LAN/SAN etc) – ie fractional ‘opportunity’ costs associated with using 200mb/s on a FC SAN link etc

  11. ianhf September 21, 2009 at 11:45 pm

    yet another one :-

    * real market value depreciation of asset over time – ie what is the expected resale market value in a given period of time

  12. Economic Truth – Gestalt IT September 25, 2009 at 7:18 am

    […] this is why the initiative started by Nick Pearce and Ian are so important. Go and read their blogs here and here on building a TCO model for storage; let’s get this thing crowd-sourced and perhaps […]

  13. […] Pearce has posted an extensive list of factors to take into account, an has promised to post a spreadsheet that can […]

  14. Jean September 28, 2009 at 3:36 pm

    Beware of disk deduplication. Some vendors boost their specs. Make sure you have a large database to compare with others.

    Tape compress very well too. They also have encryption capability for some.

    Add cost for backup time. Disk backup and cloning tape are different on that aspect.

    VTL is another area.

    I recently used to work with a TCO model tool that graphs and specs most of what you are doing. Servers, disk and tape included.

  15. Jean September 28, 2009 at 3:42 pm

    Look at PASIG here

    http://sun-pasig.ning.com/

  16. […] is keep data safe and available. When comparing different solutions one needs to ensure that the Total Cost of Ownership (TCO) of any archive system does not result in a very expensive data migration requirement further […]

  17. […] instance, the TCO model of LTO is quite a bit off based on our research on total cost of ownership. Does the report take into consideration the elements as picked out by the storage community in our […]

  18. awie October 27, 2011 at 2:19 pm

    I would like to get interested people to take part in my survey on TCO for my PHD. Anyone can mail me with their e mail address

    Thanks

Leave A Comment