After all the predictions I read about 2015 (…and many of them are pretty ridiculous indeed) I can’t help but make mine, which I would like to approach in a totally different manner, using common sense first and trying to make an anti-prediction for enterprise IT in 2015. So here is my opinion on what you won’t be seeing this year in your Datacenter.
100% Flash
It won’t happen any time soon. Many (most?) tier one workloads will be moved to flash of course, but data is adding up so quickly that it’s highly unlikely you will be seeing a 100% datacenter any time soon. It will take a few years to have about 10/20% of data stored on flash and the rest will remain on huge hard disks (cheap 10+TB hard disks will soon be broadly available for example). For mid-sized businesses, hybrid will be around for even longer than a few years, just because a different strategy is not affordable.
From my point of view, it will be much easier to see caching technology used to fill the gap between space and performance before seeing Flash surpassing that 10/20% I’m referring to. Products and technologies like the ones in the hands of Diablo technologies and software players like PernixData or Infinio (just to name a few) will be slowing down this process even further and, over time, we will see more large-scale hybrid solutions supporting fast access to the data lakes enterprises are starting to build.
And, of course, this doesn’t mean that Flash vendors won’t be growing like crazy this year!
Object storage
You won’t see a lot of object storage either (a hard thing for a fanboi like myself to say!). It’ll happen eventually, but most of the vendors will stay in their niche until they can work out a better and more solution-based strategy. I have to admit that some of them are starting to work things out very well, indeed. But they are still too small to make a dent in
the market at the moment. Data growth will however help to see them through a successful year, but we’ll be seeing much more in 2016.
Openstack
Sorry to say but I just don’t see it happening, and there are a number of reasons why which, in all cases lead to the same conclusion: complexity and immaturity. Even though the number of ties at the Openstack conferences is growing very fast, when compared to pony-tails, we haven’t reached the right balance yet. Too many projects, too many people involved (working against each other and on competing projects at times), too many vendors looking for a way to make their solution more relevant than others. The community and everything surrounding it is very active and vibrant but when it comes down to reality, real world easily reproducible use cases (or case histories) are still too few and far between to grow a true business.
On the other side you have public cloud, which is growing quite well (both in terms of features, capabilities and solutions), despite all its outages and defects. (more on this later)
Private cloud, seen as a new form of computing and not as an extension of your traditional virtual infrastructure (referring to VMware and Microsoft stacks) is only for few, and very big, end users. The complexity is just too high for most end users. And you know, even if you target some startups and all the Fortune 2000 companies (which are huge spenders indeed), how many of them are ready to embrace Openstack? How many times can that project/product/distribution be replicated without a huge/high number of work hours? And will they be able to keep up the pace with all these 6-month releases that are coming out?
Last but not least, hybrid is also getting some attention, but hard to do with Openstack. Making workloads move back and forth as well as controlling them on different Openstack clouds (which also means different releases with different APIs and features) isn’t easy. is it?
If hybrid is the “short term” solution (and short term is in quotes because short could mean years in this case), I think we will find more and more solutions based on a sort of cloud-in-a-box approach. Microsoft and HDS solutions, can avoid to write solutions?] are the first but others will surely follow. And it’s simply because if you’ve already bought something like a converged or hyperconverged solution, it’s another step further towards IT simplification. For example, a new startup like Platform 9 (founded by a bunch of former VMware employees) is working on an interesting idea, a sort of Openstack-based-management-platform-as-a-service. The product is still in pre-release but the way it’s implemented is very interesting. They are promoting it as SaaS but some of their early customers have already asked for an appliance… and I think they will eventually end up with something like that.
The end of cloud war
An easy prediction to make indeed, and this is also partly connected to what is happening with OpenStack. With Microsoft quickly becoming a cloud company (with a lot of openness which is hard to reckon in Microsoft history), with Goggle now playing more seriously and Amazon trying to keep its (unquestionable at the moment) leading position. It will be very interesting to see what happens, especially now that new interesting players (like Digital Ocean for example) are coming up. I’m not that confident we’ll be seeing VMware vCloud Air or IBM as successful as the first three this year (this is an euphemism of course) but they will have some cards to play in 2016, if they stick to their strategy.
Containers (Docker or whatever)
If 2014 was the year of hype, 2015 will be the year of maturity and 2016 will be all about adoption. Long story short, this won’t be the year for containers but most of the needed backend and enabling tools will be maturing this year.
The excitement is very high and most vendors will be supporting (if they haven’t already done so) this technology soon on their products. Not to mention how much developers love this technology!
I can’t wait to see it on Windows Server for example. With some abstraction layers, that are already present but still immature, even some traditional applications could be (partially) redesigned with minimal effort to work into a container… and when some traditional (and mission critical) applications hit the containers I think it will become a huge success. Even if they are not related at all, I won’t be surprised to see a more successful story with containers (and some specific cloud management tool for them) than with Openstack in general… it won’t happen in 2015 for sure (at least not all of this), but 2016 will see a lot of end users (even traditional ones) adopting it.
Closing the circle
I know, it’s probably easier to predict what will happen in the not too near future (and nobody will remember this post in two years, 😉 ) just because in the short term it is always difficult to find definitive and clear winners. But you know, using common sense always helps to keep your feet on the ground, even when you’re looking at the future.
Enrico some good reality and candid based perspectives, nice job. All of the items you mention IMHO will continue in 2015 (and beyond) to be relevant (as some of them have been in 2014 (or prior), however tempered with a dose of reality. Granted for different environments there may be more adoption and deployment than others.
Cheers and happy new years
Well-written and mostly in agreement. They only thing I question is Google’s commitment to the Cloud. From where I sit I see AWS, Azure, and SoftLayer making strides in corporate IT. Where is Google playing more seriously??
Larry @ NetApp
@Larry from where I sit I see Google as committed to cloud as much as those mentioned among others. However what is not seen as much is what Google is doing where AWS and others are doing a better job IMHO communicating there capabilities in a way more easily understood by the typical enterprise.
Perhaps what Google needs to do is make their cloud compute and other cloud storage offerings (as well as other services) as easy to understand and use as Google Search or Google documents vs. having to figure out what changed or how a new interface works as is the case with some of Google offerings (e.g. G+ is a good example).
I agree with Greg, I would also like to add that Google is very active w/ startups too. I know, it’s not the traditional enterprise but also AWS started with them.
Well, Mr. Signoretti didn’t go out too far on any one limb with his predictions and what he did say contains a fair amount of truth. Yes, flash won’t be taking over the world as long as it is 10X the cost of HDDs per GB. The number of workloads that need all the high IOPs you get with flash aren’t really there…yet. Flash can also be used to enhance other storage technologies like HDD, optical disc and tape. I agree that the object-based storage software market is likely to be slow and steady. No new players have entered the market in 2014 but none have left it either. Everyone expects to do more business in 2015. No OpenStack distribution has really broken out in the private cloud market. However, Red Hat is looking to leverage their RHEL customers to deliver their Red Hat Enterprise Linux OpenStack Platform. Their eNovance acquisition is designed to help them do that. I would not bet against Red Hat gaining some traction in the OpenStack market. As for the cloud wars, it looks like AWS, Google and Microsoft will constitute the major combatants. IBM (SoftLayer and OpenStack), HP (OpenStack and Eucalyptus), Rackspace (OpenStack) and others like Citrix (Apache CloudStack) will be able to carve out enough business by addressing specific markets/regions and exploiting the weaknesses of their larger competitors. More M&A activity is not out of the question either. PaaS has been the “sleeping giant” in the cloud and Docker wants to become a full PaaS provider when it grows up. We’ll see how that affects the fortunes of CloudFoundry and OpenShift, but it still looks too early to tell in 2015. So maybe 2015 will be a quiet year in the cloud.
Enrico, great observations but I agree with our HP colleague that Flash based storage will see larger adoption. If one projects historical flash pricing trends (>30%) into 2015, one can see that the OPEX cost savings quickly make SSD a winner for Datacenter in all but cold storage..
I’m curious about your thoughts on the impact of containers and Meso on ARM based processor adoption? Does this level of abstraction enable thedatacenter to easily.adopt non x86 architectures?
Bob,
Thank you for you comment.
I agree that if you consider only primary workloads flash wins.
But economics of modern hard disks can’t be beaten when you look at the overall picture.
I have clients with predictions of doubling their overall storage capacity in the next 18/24 months… And they aren’t an exception. And you know what? “Traditional primary data” (DBs, mail, ERP, crm, etc) is not growing at that pace…
ARM is very interesting tech and there is a potential in the DC for it. But I can’t see it happening in 2015. Binary compatibility is still at the base of containers mobility and developers are all on X86 machines at the moment. The number of Linux-ARM distributions is very limited too. it’s too early.