Happy new year everybody, although I intended to finish the series in 2010 but friends, snow, food and drink had priority and here I am, writing about storage on the first day of 2011.
My previous predictions were about blade servers, you can read all about it here.
As opposed to the previous post, this one is not going to be company-centric, rather I am going to take a technology-centric approach. The storage products still have some way to go before we can call them a commodity so most vendors are introducing proprietary features and functionalities in the hope that a wide enough adoption will give them a competitive edge in commoditized storage market. Even so, I see storage in 2011 broken down in 3 relevant categories, two above the line and one that is going to be mostly in testing and development in 2011 but could come up as a disruptive solution soon.
SANs
Storage Area Networks (SAN) has been the undisputed solution of choice for most enterprises for a while now but I see that changing in the near future. SANs are really fast and really powerful systems, which also made them really expensive. Pair that with the tendency of engineers to overprovision, the inability of businesses to correctly estimate their storage growth and the so-called global economic crisis and you have quite a few reasons why SANs will be falling out of fashion in 2011. This is not to say that people will stop buying SANs all of the sudden in 2011 but I truly believe that the latest and the greatest will take some time before it gets adopted in an enterprise.
Architects will have to think long and hard:
- to choose if they really need a high-end solution or the mid-range will do
- what really needs to sit on a SAN or can it sit somewhere cheaper
- what storage solutions they need for their virtual & cloud platform
Product wise I believe the usual suspects will be driving the market:
EMC
EMC has always been and will continue to be a company that markets their products aggressively with a sales and marketing team that is ahead of their R&D team. EMC has just as many followers as it has people who despise it and my feeling is that it has a lot to do with their sales tactics.
EMC offers the VMAX as their flagship product. While the product is solid and builds on quite a bit of tradition, there are few companies which really need that kind of performance (the product is high performance although EMC actively avoids SPEC). The problem with VMAX is that it costs you an arm and two legs and the remaining arm will be used to sign purchase orders for services and maintenance. According to Trefis: “Despite being commonly thought of as a “hardware” company more than 70% of EMC’s value comes from the software and services it sells to manage and optimize server hardware.” Oh, and don’t forget to look at the margins EMC is making on services.
My guess is that the VMAX will sell but very modestly in 2011. Cost conscious companies will try to get away with something cheaper.
EMC also offers the CLARiiON family. While the products are classified as mid-range, they can deliver solid performance despite some architectural shortcuts. In all fairness it’s a better choice for most enterprises and it offers a lot more connectivity flexibility than the VMAX.
I’m expecting to see good sales of the CLARiiON product line and probably EMC will invest quite a bit in bringing new features and connectivity options to their midrange product line.
HDS
HDS has come forward with innovative high performance solutions and have managed t build a legacy with their top tier SANs. Probably one of the best solutions of 2010 is the HDS Virtual Storage Platform (VSP, sold by HP as the P9500). The functionality offered by the product can satisfy most enterprises out there. The product has only been recently launched but I expect to see wide adoption in 2011 in all companies which are looking to minimize their OPEX. The sub LUN tiering, use of 2.5″ drives, reduction in power consumption and the so called 3D scaling make it a worthy proposition.
IBM
IBM’s storage division has not been able to come up with a truly innovative product since they sold their storage operation to Hitachi. The biggest problem I see with IBM’s storage is the lack of coherent and long-term vision in their product line. However IBM does manage to keep a solid market share and average growth but I suspect that this is mainly driven by their consulting and services division (exactly like in the case of the blades).
The IBM XIV will sell in 2011 but will not be amazing. My feeling is that the XIV is based on the right concept but will need some more work before it will become a true solution. I’m hoping that IBM will release another version in 2011 with improved interconnect and they really need to do something about the storage efficiency of the system.
HP
HP has their well established EVA line which they really need to revamp in 2011 if they want to remain competitive on the storage market. The EVA is simple and versatile so it will always do well with the small to medium operations but I don’t see any growth without a revamp of the family. Probably a good first step would be to come out with a 2.5″ drive shelves that would allow lower power consumption and some alignment with the top tier p9500 offer. Last but not least, integrating all the new features that HP has gained through the various acquisitions is a must.
Given the current state of affairs at HP (the political turmoil and low employee morale) there’s a big question mark if HP will really be able to maintain the focus to actually manage to integrate the all the good IP they have gained with the acquisitions of 3Par and LeftHand. However they need to come up with a homogeneous storage portfolio, the current offer is too fragmented and will hurt the company long-term if they don’t manage to provide a strategy of consolidating the solutions. It’s unlikely they will be able to deliver this in 2011 but I would be really impressed if they do.
NetApp
NetApp is one of the companies that I have high expectations from in 2011. Their long experience with file-based and clustered storage will probably help them a lot in the near future, I will write about this more in the next section.
Fortunately NetApp has managed to break free from the FUD that was surrounding them as being a low tier storage provider with file-only storage and this helped them drive decent revenue from the increased penetration in the datacenter environments. NetApp is now known to play nice with popular virtualization solutions which makes them a popular choice with companies trying out the virtualization for the first time (and there’s a lot of those) .
A word of caution is the fact that the company is a lot more present in the US and Europe than the rest of the world (at least compared to the other big names in storage). So for those running projects in the rest of the world it might not be a first choice simply due to the size of the operation in the respective area.
NAS Systems
I expect NAS systems to be the stars of 2011. According to IDC data the NAS segment was the fastest growing storage segment in 2010.
NAS systems are cheaper and simpler to manage whitch is a key contribution to a low TCO. On top of that a lot of research has gone into filesystems. Advanced file systems allow better storage efficiency compared to formatted SAN storage which again contribute to lower cost of storage. NAS systems also understand the higher level data structure which allows them to come up with faster and more efficient data protection mechanisms, freeing them form the constraints and limitations of RAID-based solutions.
Most architects have realized by now that most data is stored in files anyway. With the migration to virtual infrastructure VMs are now files and can entirely reside on NAS units. Advances in switching solutions and the migration of datacenters to 10G links make the typical problems associated with NAS systems (latency, protocol overhead) mostly go away.
I expect that in 2011 a lot more capacity will be sold in NAS units and SAN systems will be used for demanding applications like databases.
The big players in the NAS are will probably be NetApp with the ONTAP 8.0, IBM with its GPFS based SONAS product, HP with the IBIRX based X9000 and potentially EMC if it will manage to integrate the recently acquired Isilon product.
All these solutions have one thing in common: they are not just a NAS system, they are clustered solutions. This means that they allow scaling to very large capacities with little management overhead (good if you don’t really want to hire an army to manage your storage) and it allows to a certain extent the independent scaling of capacity and performance which can allow corporations to spend only what’s needed for their data storage needs.
I really expect corporate datacenters to move towards scale-out NAS based storage solutions in 2011.
Distributed Storage
Keeping your storage distributed might sound like taking a step back in your storage strategy. However if your distributed storage could be managed in a unified fashion and be part of your computing cloud that would be a step forward.
Virtual platforms are deployed now in pools of resources (memory and cpu) but are tied back to a shared storage solution in order to insure availability of the data across the pool. It’s a matter of time before we would be able to deploy storage in the same fashion, i.e. independent blocks of direct attached storage that are acting together in a unified storage pool.
Of course, it could be considered as a step back in terms of management (your storage is distributed all across the datacenter floor) but the advantages are multiple: no single points of failure and very fast access to the direct attached storage pool.
I don’t think there will be a large adoption in 2011 of this model but in the enterprise segment but certainly there are some interesting projects out there that are working on this paradigm: Ceph, Hadoop, Lustre…
Once this storage model will have wide adoption in virtualization products, enterprises will be in a much better position to deploy true private clouds based on commodity hardware.
Other companies
This post turned out to be quite a bit larger than I planned it to be but there are still some issues I did not talk about.
In short, the following companies will also have a word to say in storage in 2011: Oracle, DataDirect Networks, Symantec.
