VAST Data changes things radically

VAST Data, just went out of stealth, promises are ambitious with the plan to shake the file storage market segment with a radical new approach i.e delivering a full flash high-end NAS at a lower TCO than HDD.

The company founded in January 2016 has raised so far $80 in 2 rounds from TPG, Norwest Venture Partners, Delll Technologies Capital, Goldman Sachs and 83North. Founders came from different areas, the CEO Renen Hallak, was most recently at EMC and before at XtremIO and Jeff Denworth, VP Products, came from CTera Networks and before DDN and Shachar Fienblit, CTO, was CTO at Kaminario. Interesting we met Jeff and Shachar during the IT Press Tour in Israel in November 2016 and they both worked at that time respectively for CTera and Kaminario. So there is magic here as they claimed to be co-founders at a time they work for other companies, I probably don't get all details on the genesis of the company.

The team wishes to revolutionize the storage and breaks the storage tier approach with only 1 tier full flash. It is illustrated by the famous pyramid, shown below, with a hierarchy based on latency or capacity.

But to build such product you need some hardware components associated with key software features. Here first elements are persistent memory with Intel Optane coupled with a Flash/QLC back-end, a high speed fabric with Ethernet or Infiniband to transport the NVMe protocol, plus a wide erasure coding and aggressive reduction to finally destroy the HDD TCO. VAST Data chooses to offer the product as a highly scalable file server exposing NFS and S3.

With one tier all flash, the vision of VAST Data is to offer an universal storage. This is perfectly aligned with the concept I introduced in January 2012 with the U3 model as an Universal, Unified and Ubiquitous storage. Now the difficulty could be to run some applications on NFS or S3 and not on a block device via a disk file system. But it seems that the product will be able to run applications natively within VAST's servers and we'll get details on that in the future.

In term of architecture, the team adopts a disaggregated model that allows independent scalability for access servers and the storage servers, named Element Store. The file service runs on access servers as Docker containers, fully stateless, so very easy to deploy, to scale and move. These servers are not aware of other in the cluster and they don't need to. All data are stored on storage servers composed of Optane and QLC and addressable by any front-end server.

This architecture named DASE for DisAggregated Share Everything goes beyond classic shared-nothing model that finally limits scalability. Such implementation are good for capacity need but can't sustain the performance demand. The Single File System Image is guaranteed by the absence of cache coherency between access servers and their direct connection to storage servers that finally store all state, locks, metadata and of course data.

To extend life of components such QLC, data placement are paramount and the team spent a lot of time developing and optimizing new algorithms. And this is crucial as VAST Data commits for a 10 years warranty for QLC with a 10% support and maintenance per year.

All access servers belong to the same global unique namespace exposed via NFSv3, NFS over RDMA or S3.

The product go to market model relies on 3 options:

• full appliance for both layers access and storage,
• a mix of hw for the storage layer and software for the access service,
• and finally a pure software model with of course restrictions like size of the cluster, it has to be big, business case and deal size. When the cluster get very big, you can be flexible...

For the hardware, VAST Data sells 2U chassis with 4 nodes and 8 x 50Gb Ethernet ports. For the NVMe enclosure, it represents 675TB of flash raw, 18TB of Optane with 4 x 100Gb Eth/IB ports in a 2U chassis as well. So everything is pretty compact. Any combination can be built: 1 access chassis + 1 enclosure to deliver 30GB/s, 250,000 IOPS, 1M metadata operations per second for 1.2PB with 2:1 reduction ratio. For capacity need, you can double the enclosure, still delivering the same performance level for 2PB of capacity. And you can now double the access layer to double performance and offer now 60GB/s, 500,000 IOPS and 2M metadata operations per second.

For data services, VAST Data provides data reduction - deduplication and compression - and protection with wide erasure coding with a super high utilization ratio, snapshots, quotas, POSIX ACLs and LDAP integration.

The team did a very good job for snapshot as they're implemented as indirect-on-write mode, they are continuous, fine-grained, writable, clonable and they don't impact performance.

The product doesn't offer yet replication across cluster, SMB protocol and multiple file system so multi-tenancy could be seen as limited.

This is just brilliant, we'll see now how the market will react in term of adoption and what competition will say...

Read More

Redis Labs bags a new VC round

Redis Labs, the emerging leader in fast DB, just announced a new financial round of $60M for a total of $146M, right in time with the RedisConf early April. In the meantime, we'll have the opportunity with The IT Press Tour crew to meet and visit the company in Mountain View mid-March.

Read More

Nasuni grabs money again

Nasuni, a long time player in cloud storage gateway, is still alive despite a lack of visibility. The company founded in 2008 needs a new round of financing and just unveiled $25M for a total of $147M, wow.

With the cloud swallowing more and more data and the file storage offerings present in the cloud, the positioning of Nasuni becomes a bit fragile. Avere understood that move and got acquired by Microsoft one year ago. The two other players, Nasuni and Panzura, are getting a bit confidential and seem to lose the window of market opportunity. We'll see but again the company was founded in 2008 and is not yet profitable, it's a real sign. Is this business a profitable one with current market climate? not sure. On the other side, who cold be the acquirer? and at what value?

Read More

Griddable.io swallowed by Salesforce

Started in 2016 by Rajeev Bharadhwaj to provide data integration across hybrid clouds, Griddable.io got acquired by Salesforce very recently, it was announced January 28, 2019. Robin Purohit, CEO of the company, is already elsewhere leading Peritus.IA. Under his tenure, Griddable.io raised a Series A of $8M in February 2018 and finally concluded this sale.

Griddable.io is a SaaS platform designed for synchronized data integration across different databases. Rajeev leveraged Databus, an open source project launched at LinkedIn, as the foundational technology for Griddable.io.

As enterprises adopt aggressively the cloud with a wish to maintain a hybrid model, having the capability to synchronize data between cloud and on-premise for different databases and applications is just critical. Griddable.io creates a smart grid for enterprise data and Salesforce has recognized this need as a key wish from its clients. And with the Salesforce migration from Oracle to other database, it makes sense. As a SaaS solution, the solution runs on AWS, GCP and VMware cloud. The product uses a scale-out grid architecture with a policy engine to provide a mechanism to synchronize transactions with sub-second latency with the guarantee of the order of transactions. We'll see if Salesforce will offer this as standalone product or use it internally or even both.

Rajeev was the founder of Ejasent in 1999, pretty famous for UpScale and MicroMeasure, acquired by VERITAS Software for $59M in cash in 2004. At that time, Ejasent was led by a champion of the sale of companies Jason Donahue. Just check where he's today and you know the future of the company he leads. That acquisition by VERITAS help the company to promote its Utility Computing model that anticipated the Cloud.

Read More

Full Flash NAS, the battle is on

Dell Technologies announced yesterday a new iteration of their scale-out NAS offering with again a full flash model, the Dell EMC Isilon F810. This model is a 4U chassis grouping 4 nodes, 60 SAS/SATA SSD so 15 per node from 230 to 924TB raw per chassis. The performance per chassis are 250,000 IOPS and 15GB/s. The company claims to deliver 9M IOPS and 540GB/S aggregated throughput for a full cluster. In the Isilon terminology a cluster starts with 4 nodes - here 1 chassis - up to 144 nodes or 36 chassis. This limit exists for years and doesn't represent a real issue with the current density of HDD, SSD and server.

The solution is very rich in term of features but costly making real differences against competition with additional data services such inline data Compression up to 3:1 and deduplication with SmartDedupe, Replication with SyncIQ, Tiering with SmartPools and CloudPools, Compliance with SmartLock and FlexProtect for erasure coding. Many competitors don't even mention these features with a specific name as they bundle them as standards functions.

At the same time, Dell EMC mentions its new gadget, ClarityNow, a data management software acquired last summer from DataFrameworks. In fact, they swallowed the company without any public announcement except a blog post. I wrote about it in August. With that move, Isilon confirms its lack in analytics and having ClarityNow doesn't change anything, the software is just an add-on, even its GUI is still the ugly one DataFrameworks used since the beginning. The support is still the same file servers and object storage and of course now Isilon NAS and ECS family. Two questions remain open: why Dell didn't build its own analytics layer? Why did they pick this toy?

We pick a direct competitor of Isilon, Qumulo, and we detail the configuration with a cluster of 4 nodes. Ok we can extend this approach with same rack unit, capacity or performance comparisons, we let you do that exercise.

Comparison of 4 nodes cluster

Configuration	Isilon F810 cluster	Qumulo 92T cluster
4 nodes cluster	1 chassis 4U (4 nodes)	4 chassis 8U (4 x 2U node)
Storage	60 SAS/SATA SSD (15 per node, from 3.84 to 15.36TB each)	96 NVMe SSD (24 per node, 3.84TB each)
Capacity	230 to 924TB raw	368TB raw
Performance	250,000 IOPS 15GB/s	400,000 IOPS 16GB/s

The battle is on with other players as well such Panasas and Weka and of course specialized flavors based on Spectrum Scale, Lustre or BeeGFS even if some of them have serous limitations.

Read More

Lima is dead, a lesson for founders and investors

Unknown for the market, the market gem Lima as some people named it, just closed its doors and it's not a surprise. The opposite was a miracle. The facts speak for themselves, there was no market, despite a funds raised from Partech Partners following a huge Kickstarter campaign well above expectations. It confirms once again that the metric of crowd funding is wrong and doesn't validate anything and say that Lima was a gem is purely a joke and a big mistake based on this Kickstarted amount. Partech confirms also its difficulty to pick and select data storage candidate, is it a lack of analysis or expertise? probably but above all prove that France is absent of this sector. As Lima's founder said, the story was a miracle with 80,000 users, and a miracle das some difficulties to last. I say for a long time that the market is always right, some people forget this promoting disruption... It illustrated again the reign of clouds giants that remove all obstacles if we can consider Lima as an obstacle. But in Lima case, I thought the good choice was to integrate the logic within home NAS, having a separate toy didn't help, was seen as a gadget and finally miss its original mission. A good lesson.

Read More

Qumulo introduces Data Migration Service

Qumulo, the leader in high performance NAS for enterprise demanding applications recently unveiled a partnership with Atempo to boost the data movement towards their file storage infrastructure. The Qumlo offering base don Atempo Miria is Data Migration Service aka DMS grouping a software and a practice.

Atempo Miria seems to be the new name of Digital Archive aka ADA and was chosen by Huwaei with a deal announced in October in Shanghai during the Huwaei Connect conference and more recently with DDN shared at SC18 under the name DataFlow. I invite you the ready to check this post.

We checked the web site of Atempo and realize the company is behind the announcement as the product page doesn't mention Miria and Qumulo is absent of the partner page as well.

Qumulo DMS has several interesting characteristics: dedicated HW server(s), jobs parallelism, linear scalability with data movers addition, no downtime, incremental capabilities and security schemas preservation.

The service is available through Qumulo channels. Who will be the next Atempo partners after Huawei, DDN and Qumulo? WekaIO, Panasas...

Read More

LogicMonitor accelerates with partner program

LogicMonitor, a leading SaaS-based performance monitoring platform player, announced a partner program composed of resellers, systems integrators, MSPs and technology integrators.

Following a great 2018, the company accelerates and plans an real expansion in 2019 in a a $47B IT infrastructure worldwide.

Met last October with The IT Press Tour, the company has impressed the press team with a very comprehensive product. The company so far has more ten 30 press articles following the meeting.

Read More

Sylabs shakes the palm

During the recent session at RStor last December during the 29th IT Press Tour, we had the privilege to spend time with Gregory Kurtzer, the CEO of Sylabs, an alternative container player. Greg is also the creator of CentOS and Warewulf projects.

Sylabs delivers the dream of every IT admin to run any application anywhere and anywhere means private or public clouds or edfe and event IoT platforms. Even if de facto standards exist such Docker, Sylabs changes the packaging and modify paradigm with Singularity.

Open source with a Community edition, Singularity becomes the container of choice for HPC, EPC/AI (Enterprise Performance Computing) and Cyber Security thanks to an huge installed base, millions of cores running and support of GPUs, key component in scientific and technical computing. The product spans enterprise, research labs, clouds and universities, and it also boosts GPUs enterprise penetration, it is faster than Docker, more simple and above all it is universal.

Designed with security, mobility and performance in mind, Singularity has 2 differences:

1. the presence of a runtime engine with OCI compatibility and work it Kubernetes, Mesos, Kubeflow and Nomad.
2. the container format, aka Singularity Image Format or SIF, with only 1 signed, immutable and encrypted file.

The security is also a key advantage in favor of Singularity if you compare with Docker. There is no root owned demon process and containers don't run as root.

As AI, Data Analytics and HPC become more and more common within the enterprise, meaning that players like Google, Microsoft, Red Hat, HPE, Dell or NVidia leverage pSingularity to be platform agnostic, minimize risk and reach larger audience.

In addition to the Singularity Community edition, the company has developed different flavors: SingularityPRO, Singularity Cloud and Singularity Enterprise. The Pro offers subscription, licensing, professional services and support. The Cloud flavor, in alpha today with plan to be GA in February, provides a container library, build services, keystore and a devops platform that ties all these pieces together. The Enterprise edition groups the PRO and the Cloud to run on-premises.

Sylabs plans to offer also SyOS, a thin micro-OS appliance to secure and control containers on any x86, ARM and GPU platforms, and DCM aka Distributed Container Management for on-prem, cloud edge and IoT deployments. Both of them should be released soon probably in February or March.

A real good surprise met a few weeks ago and we'll follow closely this story and it is very compelling and shakes market established positions.

Read More

Silently, Minio eats the object storage world

Minio continues to expand its world domination. The company launched in 2014 by Anand Babu Periasamy, Garima Kapoor and Harshavardhana, became the de-facto standard for everything related to on-premises S3 storage.

Partners and users realizes that and multiplied integration in various products combinations, demonstrating the flexibility of the product. The other players initiated before them started to disappear from the market landscape, being replaced by Minio, or even have chosen to do something else. In other worlds their opportunity window disappeared. Among object storage vendors outside the cloud, there are just a few that count, the 20 other became anecdotic. I remembered some vendors trying to compete with them claiming superior technology... haha what a joke especially from leaders who don't understand storage, trends, market forces and technology moves.

Datera and iXsystems have chosen Minio and I won't be surprised to see others doing similar move. It can be deployed with Kubernetes, Docker or Mesos, even on NAS and offer an S3 interface for GCP and Azure. Even some competitors use their mc client on top of their product.

Recent meeting with the team has shown that they address new use cases not very well touched by their competition or object storage in general. I mean here big data and analytics, machine learning and deep learning and more generally application data. We see partnerships with IBM, Azure, Pivotal, Vertica, Greeplum, NetApp, Intel, Samsung even Dell EMC. The product already supports Lenovo SR650, Cisco UCS C240, Dell R470xd and HPE ProLiant DL380.

They also use S3 Select which gives a new dimension to S3 storage and a new life for application integration, this is so powerful, thank you Amazon.


I already said that the Minio Tsunami is coming but in fact it already started as I wrote in December 2016... But we know that Minio won't do an IPO but will be acquired by someone, who, when and how much? Today I estimate the value between $500 to $700M to start to discuss.

Read More