May 25, 2016

New blood in storage with Symbolic IO

Symbolic IO (, founded in 2012 by Brian Ignomirello, former CTO of HP global storage team, is now out of stealth. The company claims to be the first computational defined storage solution. The company introduces IRIS, for Intensified RAM Intelligent Server, with the goal to be media-agnostic and data driven. IRIS is delivered via 3 flavors: IRIS Compute, IRIS Vault and IRIS Store available as 2U units powered by Intel cpus. The team has built a radical new approach for storing data able to reduce the storage footprint in a great amplitude promoting a super dense 2U and be super fast working with RAM with non volatile entities. When we mean super dense storage, we didn't imagine that 1PB can be stored in just 2U. Without telling us exactly how it works, of course, Symbolic IO is about a new way to provide Data Virtualization with what the company says a "materializing" and "dematerializing" data in real-time, living in memory without any reality on storage.
In term of performance, Brian Ignomirello claims that IRIS is 10x faster than 3D XPoint and 10,000x faster than NAND flash. Wow, impressive and probably unbeatable at the time of the product release in Q4 2016. The company has worked super hard to develop OS, ASIC and algorithms for 4 years to reach that super level of performance. The product should appear on the market before the end of the year and will start at $80,000. From what we know today, IRIS is pretty unique and it deserves a look and test, no doubt. Symbolic IO would be a perfect candidate for a future edition of The IT Press Tour.

May 24, 2016

Datameer New Generation of Analytics

Datameer (, one of the leaders in data analytics preparation and a good surprise of the 18th IT Press Tour in March 2016, just introduced the 6.0 release. This major product iteration offers 3 new key elements:
  • a new user interface to radically change how users interact with the platform,
  • an extension to Datamer's Smart Execution with the support of Spark processing engine that simplifies data mining and machine learning,
  • and a continuous data discovery mechanism.
The outcome resides in users behavior with visualization at each step of the process, the workflow is now more fluid. No need to wait the final phase to obtain result and finally discover some mismatches or inefficient sub-phases. Each step feeds the process and offers visualization. The goal of this 6.0 is to simplify, reduce processing time and improve the user experience. Battle is on for data preparation and Datameer clearly occupies a leading position with a limited number of other players. Congratulations to the team.

May 23, 2016

AtScale raised a new round

AtScale (, a leader in BI on Hadoop and a new company met during the 17th IT Press Tour, announced recently a series B at $11M. The total raised now touches $20M, which is interesting and at the same time not so important to accelerate the growth. It seems that the team wish to keep the control and their destiny and as an effet, could reduce the growth speed versus competition. What is sure is AtScale delivers a rapid growth during last 12 months with 5 fold revenue expansion - ok was very tiny - and adopted by famous customers and brands. Among various investors, Comcast Ventures has joined the pack and Amr Awadallah, founder and CTO of Cloudera, is now a board member. Congrats to all the team and we expect to see you again for a future edition of The IT Press Tour.

May 19, 2016

Primary Data extends the data lake

Primary Data (, the Data Virtualization company extending pNFS, just announced new targets as data servers within the data lake built and controlled by its DataSphere software. With this new release Primary Data supports now EMC ScaleIO, EMC Isilon and Amazon S3. The company is able to support multiple back-end - block, file and object - local or remote - and unify them within a unique namespace. For some people who wish some background on Primary Data, I invite to read a post I wrote 2 years ago available here. Good progress.

May 16, 2016

When object storage promoters realize the role of files

It's fun, really fun, to see in some videos, posted on YouTube taken at OpenStack Summit in Austin last month, SwiftStack CPO Joe Arnold recognizes that Object APIs are for developers and finally accepts the reality of the high and ubiquity presence of file-based application especially commercial ones that don't speak object at all. And now these guys that serve as Ambassadors of Object stuff promote file access via ProxyFS, wow, what a conversion. They use to refuse file access, then they embedded Maldivica gateway as the market asked for such access and now they speak about native file access. In fact, there is nothing in the market, ho sorry yes it's new for Swift, ProxyFS allows file and object access to the same content. Again, really a new feature for Swift but pretty common in the industry by other object storage. I love these videos and if you don't find them, here they are:

May 11, 2016

Evan Powell is back in storage

Evan Powell, the long time CEO of Nexenta, is back as a board advisor of Cloudian. He will join his past colleague Jon Ash, who was VP of Sales at Nexenta, now running WW sales organisation at Cloudian. Good news, we'll probably have the opportunity to meet him again during the June IT Press Tour as Cloudian is a second time sponsor. To refresh memories, Nexenta has participated to the real first edition of the tour in June 2010.

May 10, 2016

E8 Storage attracts 12 more millions

E8 Storage (, an innovator in Flash-based storage for the entreprise, just announced a Series B with $12M from Accel, Magma Venture Partners and Vertex Ventures. The current total amount raised reaches $18M.
The storage industry really lives a revolution with a few technologies and especially everything related for Flash. E8 Storage belongs to the second wave of players leveraging NVMe huge potential to design and build new offering and to address new applications and business challenges.
The company has developed a radical new approach with a shared NVMe storage appliance able to deliver 10x the performance of existing market AFA, 10M IOPS and very low latency pretty similar to local NVMe devices at a range around 100µs/50µs of read/write. Typical applications could be a trading floor with high frequency trading needs that require high speed and low latency storage back-end. The expertise of E8 Storage resides in storage software for demanding environments and here it translates in performance consistency and guarantee - no need for caching -, data protection with double parity model, access coordination to the shared device and auto balance control via an advanced multi-pathing layer. In term of SSDs, the appliance embeds very standards Intel NVMe SSDs. A joint paper with Intel and E8 Storage was written recently and is available here.

The rest of 2016 will be interested with new players such Mangstor, Apeiron Data Sytems, DSSD (acquired by EMC in May 2014), FlashGrid, Excelero and of course E8 Storage, competing against each other but also against more classic vendors, strange to say, but AFA vendors are now traditional stuff. As a technology it's very interesting to see that NVMe generates 2 different storage architecture approaches: first, a central storage appliance star topology with Apeiron, DDSD, E8 or Mangstor and second, Excelero or FlashGrid where every NVMe SSD is connected locally on edge node building a mesh or peer-to-peer architecture with a shared logical pool. Always good to see new players and new ideas to bring new value and shake the established storage industry palm. And E8 Storage will participate to the 19th edition of The IT Press Tour in June. Super.

May 9, 2016

IT Press Tour #19 - 6 years later

The IT Press Tour (, launched in June 2010, will celebrate a team success for the 19th edition next June. So far, the tour has met 137 companies, delivered 200 sessions as some companies did multiple editions (for example, Cloudera did 5 editions, Nimble Storage, Pure Storage, Solidfire or Hortonworks, 4 editions...) and generated more than 5000 articles in online and print magazine. The tour has largely contributed to create visibility for invisible companies, its success speaks for itself.
The next edition end of June will be dedicated to IT Infrastructure and will take place in San Francisco and the valley. As usual, the tour team has invited many innovative companies, many of them are pioneers in their respective segment, and the list as of today is:
  • Cloudian, the best object storage for on-premise S3,
  • Datrium, the premier supplier of server-powered storage systems as they name themselves,
  • Hedvig, a new SDS approach by the father of Amazon Dynamo and Apache Cassandra,
  • Kaminario, one of the fastest All-Flash Array,
  • OpenIO, one of the few pioneers of object storage,
  • Portworx, a leader in software-defined storage for containerized applications,
  • Rozo Systems, a super fast scale-out NAS,
  • Springpath, innovator in hyperconvergence software,
  • Sysdig, first and only comprehensive set of container-native visibility solutions,
  • Versity, key player in data tiering for demanding environments,
  • and some confidential ones we can't mention today.
Follow the tour with the hashtag #ITPT, you can also follow dedicated Twitter handle @ITPressTour and other press group members. This tour will be exceptional... again... for everything related to storage.

May 5, 2016

Excelero will take off very soon

Excelero (, innovator in scale-out block storage, will unveil officially the solution very soon. The web site is still empty so the info is not very available. I had the privilege to attend MSST 2016 and Josh Goldenhar, VP Products, and Yaniv Romem, CTO and co-founder, were there to present and speak about the Excelero approach. The company was founded 2 years ago, received $20M from Battery Ventures and has of today 25 people. David Flynn, CEO and Founder of Primary Data, has also invested in the company and is a board advisor. Excelero develops a very promising product named NVMesh, a pure software product that leverages, NVMe SSD, SR-IOV and RDMA. The philosophy, still a shared-nothing one, offered a decentralized server-based super fast flash-based block storage. NVMe SSDs are connected locally to individual servers and then connected via a super fast network. The result is a very fast shared block storage pool available from every host. The first results have demonstrated super results 20 to 30 times better than "classic" AFA. NVMesh allows remote NVMe devices to operate like local devices under 100 microseconds. In details NVMesh has 3 components:
  1. a client block driver running on each participating node who needs and uses NVMesh logical block volumes,
  2. a target module for NVMe target drives identification linked to the management module,
  3. and a storage management module to offer centralized management via WebUI and RESTful API.

In term of features, NVMesh can built RAID 0, RAID 1/10 volumes and "wide striping" across multiple NVMe SSDs connected to other participating servers. Availability is reached by multi-pathing and mirrored volumes. Today 128 targets can participate per NVMesh instance. For performance, the product is very fast with 4M read and 2.8M write 4k IOPS, 16GB/s throughput and super low latency with 90µs/25µs for read and write from client to server.
A pretty interesting data sheet with Mellanox and Supermicro was presented recently at the OpenStack Summit, with Cinder integration and is available here. It shows what is probably one of the fastest block implementation for OpenStack. Yaniv Romem, CTO of Excelero, has presented the approach also during the Flash Memory Summit 2015 and I have extracted 2 interesting slides:

Excelero has demonstrated a super fast, robust, efficient and none intrusive product with great simplicity. It will be introduce in July and the pricing model should be based on number of drives and servers but independent of the capacity.
The product reminds me what FlashGrid is doing, yes dedicated to Oracle world but pretty similar philosophy. I invited you to read the post I wrote about FlashGrid in October 2015. Next months will be super interested. Excelero belongs to the small group of 2016 storage gems.

May 3, 2016

SwiftStack 4.0, interesting release

SwiftStack (, clearly the leader in OpenStack Swift practice, just announced the version 4.0 of their distribution. Among the list of features, we see the following ones:
  • An integrated Load Balancing capability, the argument used is base don the cost of external LB. We can understand that but many companies already have them deployed so there is no additional cost. The argument works only for people who don't have yet any LB.
  • MetaData search, hum, it reminds me what DDN has introduced with WOS 3.0 in September 2013. The difference here is the connection with ElasticSearch which is a good point.
  • Support of IPv6,
  • SwiftStack Drive as an optional desktop client. Wow what a mistake, I hope and imagine it took many long internal discussions to justify this approach trying to sell an end-to-end solution. And guess what, the partners who offer this kind of tool has now a competitive offering from the back-end vendor itself. It reminds me something... It let you imagine what are their reaction. No problem, we all learn from mistakes so to make progress you have to do some...
So what are the new things, does it deserve a major 4.0 release ? Wow sorry it was OpenStack Summit and some players has to introduce some new things even little ones...