Death of the SAN?

Recently market analyst Wikibon predicted the death of the traditional SAN market in the next 10 years. The resuts of the report were summarised in The Register with the headline news being that existing SAN, NAS and DAS revenues will drop to almost 90% of their current levels in 10 years. So what is causing this change in the fortune of traditional SANs, hyper-converged storage says Wikibon. Hyper-converged storage is a solution that pools storage from inside servers and presents it out as shared storage.

It’s well documented that the traditional SAN’s revenues are being squeezed by the cloud, and newer solutions to storage such as hyper-converged and software defined.  However, the dominance of hyper-converged that is being predicted just doesn’t make sense let’s look at why.

All or Nothing

We technical people love to see things in absolutes is it on or off, fast or slow a one or a zero.  However in reality life is not like that, there are in-betweens or ‘grey areas’. This is the same with product choice rarely will a product be the right choice for almost everyone, there will be certain circumstances and use cases it is more adapted to than others.
The marketing message from the hyper-converged players has been no SAN, again back to the IT style of all or nothing thinking. But a product type claiming to be the fix-all solution to all problems sets alarm bells ringing to me.  I get visions of the latest fitness gadgets you see on selling channels: “The arm-blaster 3000 the only fitness device you will ever need, works the whole body, relaxes the mind, and makes you a cup of tea at the same time.” As much as the marketing men would like us to believe it there is no fit all solution for any product in any industry, it’s about choice and use cases.

bull worker

A compelling case

A product needs to be truly compelling to get people to get people to leave their exiting solution and move across to another. Virtualisation was an example of one of the most compelling cases for changing technology in the last few years and possibly ever. Who wouldn’t want to reduce their datacentre foot print in turn reducing cooling/ power costs, obtaining greater flexibility and agility with greater uptime? Being a salesman for VMware during those days of mass conversion from physical to virtual must have been a dream, as the customer really had to answer the question why wouldn’t we do this?

I do not believe that the use case to swap to hyper-converged is anywhere near as strong. Will this improve performance, availability and reduce costs for every customer like VMware did? Unlikely. It will certainly have use cases where hyper-converged will be a better choice but not so broadly across the market that it will end up accounting for 90% of the market as Wikibon predict.

no san


What any report that aims to predict into the next decade fails to take account of is innovations that occur during that period and can be game changers. These are often difficult to spot even when you are a key play such as Bill Gates almost missing the importance of the internet and Nokia completely missing smartphones. As yet we do not know how the development of other storage technologies like Object based and cloud computing will shape the future. Add to this true innovations in the pipe line like HP’s memristor and other memory based storage technologies under development and things could be more different in 10 years than any of us could guess.

Closeup of HP's Memristor devices on a 300mm wafer.

World Peace

Well world peace may be a little ambitious but I think it’s fair to say traditional SAN’s should be able to co-exist and indeed benefit from the development of hyper-converged and other new storage technologies. Rather than all sides claiming their product is the only way I would like to see developments that allow the two to co-exist with each other. SANs and hyper-converged have different use cases and are not mutually exclusive, so value is to be added to the customer by developing co-existence between the two. Co-existence could for example be traditional SAN and hyper-converged using the same management tools or data replication between the two, this would make more sense for the customer rather than having to chose between the technologies. Software defined storage was supposed to move us away from islands of storage, not create them by drawing even deeper lines in the sand.



The free market at its core is about competition, choice and innovation.  Never has the storage industry been more competitive with the era of SSD’s and software defined bringing many new entrants each with a different approach. Hyper-convergence is one of the big success stories coming out of this new era of storage and it absolutely appears to be here to stay and will continue to grow its market share. But I do not foresee it achieving near market saturation, as that would bring very little benefit to the consumer, long live innovation and choice.


To stay in touch with more content connect with me on LinkedIn and Twitter.

HP 3Par Replacing a Failed Disk

Replacing a failed disk in a 3PAR is pretty simple you just need to follow a few steps to make sure you do it safely. If you are new to 3PAR or would like to learn more a good place to start is our 3PAR beginners guide

Let’s get started with the disk replacement procedure:

1 Check to see if you have any failed or degraded disks in the system. Take a note of the disk ID and cage position. In this case disk ID =26, cage position = 2:8:0

3PARSAN01 cli% showpd -failed -degraded

                           -Size(MB)-- ----Ports----

Id CagePos Type RPM State   Total Free A     B     Cap(GB)

46 2:8:0? FC   10 failed 417792   0 ----- -----     450


 1 total                  417792   0

2 Check if the disk sevicemag command is running on the drive. The servicemag command is used to inform the system to evacuate all the chunklets from a drive so that it is ready for service. Below we can see the servicemag has succeeded on the drive we identified in step 1.

3PARSAN01 cli% servicemag status

Cage 2, magazine 8:

The magazine was successfully brought offline by a servicemag start command.

The command completed Thu Jul 10 20:07:03 2014.

servicemag start -pdid 46 – Succeeded

3 Next we double check there is no data left on the drive. You can do this by running showpd –space driveID as below. You need to check that all columns other than size and failed are zero

3PARSAN01 cli% showpd –space 46


Id CagePos Type -State-   Size Volume Spare Free Unavail Failed

46 2:8:0? FC   failed 417792     0                 0     0          0         417792


1 total                        417792     0            0   0          0        417792


4 Next to replace the physical disk. Make sure you are happy with the above steps. Then pop that bad boy out, you will have a note of the location of the failed drive from step 1.


5 Once the disk is in you can monitor the progress of the rebuild by running servicemag status, which will give you an ETA for completion.

3PARSAN01 cli% servicemag status

Cage 2, magazine 8:

The magazine is being brought online due to a servicemag resume.

The last status update was at Thu Jun 26 12:09:19 2014.

Chunklets relocated: 73 in 50 minutes and 34 seconds

Chunklets remaining: 400

Chunklets marked for moving: 400

Estimated time for relocation completion based on 41 seconds per chunklet is: 4hours, 57 minutes and 39 seconds

servicemag resume 2 8 -- is in Progress

6 Once this is complete you can check that your disks are showing in a normal state with showpd -state

3PARSAN01 cli% showpd -state

Id CagePos Type -State- --------------------Detailed_State---------------------

43 2:5:0   FC   normal normal

44 2:6:0   FC   normal normal

45 2:7:0   FC   normal normal

46 2:8:0 FC   normal normal

47 2:9:0   FC   normal normal

That’s it, job done! This blog has over 150+ 3PAR articles we have put together a selection of the best articles for you to learn more about 3PAR


To stay in touch with more 3PAR news and tips connect with me on LinkedIn and Twitter.