Top Storage Trends for 2013Of course SSD and cloud, maybe big data
By Jean-Jacques Maleval on 2013.02.06
As usual and since several decades, the amount of data continues to increase and this movement is not going to stop. In parallel the price/GB is decreasing. Nothing new. But, following the Thai flood, the cost of the crucial element of a storage system, the disk drive, increased drastically. This bad event is now far from now but the prices of the HDDs didn't came back to what they were before the flood. It was good news for SSDs especially as their own prices continue to decline regularly. For sure, the price/GB of SSD will continue to be higher for several years, but the trend is not in favor of HDDs, having difficulties to increase in areal density. Progressively storage systems are going to use more and more SSD as tier 1 or primary storage, with HDD, being progressively a backup media, for tier 2.
What will be really the killer trend for 2013?
Already twenty companies in the world have designed storage systems for SSDs only but no storage giant is among them - with the exception of IBM following its acquisition of Texas Memory Systems for high end and very expansive units, and maybe HP with 3par. One of the biggest event of the year will be the announcement by EMC of its XtremIO, following the acquisition of an Israel-based start-up with the same name.
We don't say that this product will be revolutionary compared to the current offerings from other companies already there. But it will give to all-SSD systems the recognition they need with the arrival in this field of the ≠1 storage company to reassure the end users, notably the big ones that don't want to deal with small firms or start-ups for a device that will manage their critical applications (database, data warehouse, big data analytics, storage for lot of VMs, etc.) without the giants' quality of services. What will probably be missing is the possibility to integrate these SSD arrays into the current users' global storage configuration. EMC is not supposed to do it with the first version of XtremIO.
We expect acquisitions in this field by big storage companies who absolutely need to rapidly get into this market.
SSDs progressively killing HDDs
SSD are progressively replacing HDDs. It's already the case in all portable phones, smartphones, e-books, tablets, netbooks, etc., all the devices smaller than notebooks. For these latter, we expect no more HDDs into them at all in the next five years. Then they will be integrated in standard configuration of PCs.
Unfortunately for HDD makers, they are unable to increase capacity at the speed of flash chips, continues to diminish in term of nanometers and the arrival of TLC, the two parameters authorizing the reduce prices of flash-based devices.
There is no more disk drives smaller than 2.5-inch and, in this form factor, some companies are already offering - expansive - flash units at 2TB and even more, the capacity limit of today's disk drives in this volume.
Tiny eMMCs are a dream for designers of small computer devices with today's capacity as high as 64GB.
SSDs have also progressively entered into data centers together with SAS HDDs, but the arrival of all-SSD systems will change the way big computers and HPC will be used, notably with PCIe SSDs.
USB keys will pursue their success, once more with their increasing capacity soon able to replace lower-end external HDDs.
Among the trends, we cannot miss cloud storage, not really new but booming. This market will expand greatly when the Internet bandwidth will be accelerated.
But how the 2000+ companies offering online backup in the world can be profitable is a mystery. They try to compete adding functionalities and mainly by offering more gigabytes for free.
Hybrid backup, onsite and public cloud-based offsite, is an excellent way for data protection, but we see bigger opportunities for more secure private clouds. Some companies will never use public cloud for confidentiality reasons.
These two words mean nothing. A data can be big, and then? Big data analytics is more understandable.
When is it going to happen? Storage analysts and their biggest customers are crazy about this new technology. Can you believe in these figures published by Gartner last October: big data will drive $28 billion of IT spending in 2012 and $34 billion in 2013.
You cannot see a piece of marketing communication in the industry without these two words. Remember that it was the same buzzword few years ago with ILM (Information Lifecycle Management) excessively pushed by EMC. Not any more.
In reality, there is just a bunch of companies who have adopted big data analytics even if many of them are looking at it. It could be a paradigm for the storage vendors because they will need more capacity to handle it but not necessarily adding a complete new storage system.
Apache Hadoop is the dominant big data analytics platform.
The tape market is declining since several years. The sequential media is no more in favor for backup, being to slow compared to HDD. And de-dupe is not offered for tape.
Nevertheless, tape will not die. Its last niche is for long term archiving of enormous capacities - with access time in minutes and even in hours - in huge libraries, the only application where its price/GB is lower than disks. Recent installations by Amazon and Goggle are confirming this trend.
FC is progressively losing market share against cheaper Ethernet and faster IB but 16Gb FC is promising. But just a proof: there is no more HDDs with FC interface. 12Gb SAS is supposed to come later this year.
Other trends in 2013
- Software-defined storage
- Growth of unstructured data
- Object storage
- Cluster storage
And coming of age
Some other subjects we already predicted in January 2012 for last year: storage virtualization, scale-out NAS rather than SAN, Thunderbolt or USB 3.0 (4.0 to come), data protection of notebooks in enterprises, HPC storage. They are coming of age.