Exclusive Interview With Zahid Hussain, CEO and Co-Founder, Vexata

He is:

• Co-founder and CEO of Vexata Inc.

He was:

• SVP/GM at EMC Corp.
• VP engineering at VMware Inc.
• EIR at August Capital
• VP engineering at Brocade Communications Systems Inc.
• VP engineering at CoSine Communications Inc.
• Chief engineer at SGI Corp.

StorageNewsletter.com: Could you tell us the story of the company and the root of the project with the background of founders as it is a key piece?
Zahid Hussain: Surya Varanasi and I both come from deep networking, storage and virtualization backgrounds. Our purview running key businesses at VMware, EMC and Brocade has given us a unique appreciation of how the markets and enterprise data centers have evolved over the course of the last decade. What we noticed was a compelling confluence of core elements like compute, networking and storage media were progressing at a rapid pace. We also realized that the compute and networking ecosystems had done a good job of matching to the growing application needs in the market. The laggard in the space were the storage architectures which had ridden the benefits for media evolution, but had not really changed the architectural approach to managing and consuming the next generation solid state media. This meant that as the enterprises were emerging into more real time ecosystems where data access at scale was a critical element, the storage architectures were providing nothing but compromise. We felt that if we combined some of the core principles we had perfected in the past into the next generation of data infrastructure, we could create a truly differentiated offering that is matched to the emerging needs of today’s highly scalable application environments. It was a unique opportunity for us to bring our networking, virtualization and storage experiences into a ground breaking infrastructure that could address where the enterprise application ecosystems were headed. So in the early part of 2014, Surya and I started Vexata to bring to market a new kind of data infrastructure that would address the pressing application challenges of real time data access at scale, really closing the performance and economics gap.

Why do you think the market needs an other all-flash storage array, the market being saturated?
It is not so much that the market needs another all-flash storage array, but rather the fact that enterprises need a way to address their application performance and scaling challenges. Enterprises want simplicity without compromising on performance, economics and scale. We often hear from our customers that archane storage architectures are the biggest culprit in the ability of enterprises to meet the growing application data access challenges. These challenges could be existing platforms based on Oracle, SQL, etc or emerging newer workloads in the IoT, AI or machine learning segments. In either case, customers need a simple way to deliver to the performance needs of these applications. The fundamental reason behind this is that most AFAs introduce latency which prevents them from taking advantage of NVMe media, so any advancements made in silicon do not translate into improved system level performance gains. By removing latency from the I/O stack, Vexata has completely changed the landscape and can fully exploit performance advancements made in SSD media. Vexata has built a system that is completely differentiated as proven by many of our customers during direct comparisons between Vexata and other all flash arrays. In some cases, the performance advantages delivered by Vexata were so compelling that they deployed Vexata in place of their existing array which is also a testimonial to the plug and play nature of the Vexata solution.

So far Vexata raised $54 milllion, any new round in 2018? Are you far to be profitable?
Yes, we’ve raised $54 million to date and with top-tier investors such as Mayfield, Lightspeed, Redline and Intel Capital. The caliber of our investor portfolio is a testament to what we are seeking to achieve and each of our investors provide a unique perspective, but they all view Vexata as a strategic part of their portfolio that can really shape the future of the data storage market. Aside from that, I can’t really comment any further on funding at this point.

VX-100 is a block all-flash storage array, what’s the philosophy of the product?
Our product vision from the beginning has been to take a systems level approach to delivering high performance data access without having to disrupt how enterprises deploy these systems or requiring proprietary software loaded on application hosts. The root of the problem that we solved was to completely remove I/O wait times to deliver optimal performance without changing the stack. Our core architecture is built upon the Vexata Operating System (VX-OS) which allows us to utilize the best of breed, commercially available hardware that works in conjunction with our software and management to deliver a system that delivers amazing performance, extremely high reliability at a very compelling price point. This unmatched combination of performance and economics completely changes the game. While natively we offer block access, we also provide a full compliment of file offerings to support, scale-out file, NFS, SMB and object.

Today VX-100 takes a scale-up approach, but what a stacking approach of several of them to offer a horizontal model?
Yes, the VX-100 provides in-box scaling allowing users to start with as few as 3 ESMs (Enterprise Storage Modules) providing 25TB of usable capacity, but can scale a loaded chassis of 16 ESMs for 180TB of usable capacity. Because each ESM contains data path acceleration, performance scaling is very linear with in the chassis by fully utilizing the internal data paths with massive parallel processing and over 256GB of full-duplex bandwidth available in the system.

VX-100 occupies on 6U in a rack. A deployment of 6 VX-100’s in a rack will give the user 1PB+ of storage that can be accessed at 300+GB/s. Servers connected to the storage fabric will be able to provision volumes from across the pools of VX-100 and will be able to access data through the active fabric at over 300GB/s.

What are the key differentiators? We heard it’s super fast?
Yes, as I referenced, the key design objective was to deliver a system that delivered this amazing performance, but to also simplify operations and reduce costs. Because of our architecture, Vexata can deliver the raw performance from 12 NVMe SSDs that may require 50 or 100 SSDs from other AFAs. Our goal was to simply deliver the full potential of the underlying media to the application. This efficiency completely changes the economics because the architecture fully utilizes the performance of the SSDs, unlike other AFAs that waste the media performance. With existing AFA architectures, you’re paying a premium for these drives and they offer nothing more than spare capacity. Not a winning economic formula.

Let’s take the VX-100, it’s an enterprise class system that fully leverages the power of NVMe media, but also has the performance to leverage technologies such as 3D XPoint. The performance has been verified across a number of independent benchmarks, including a recent ESG benchmark report. These are proof points that demonstrate the ability to fully utilize all (16) 32Gb FC links, which simplifies deployments across a broad number of use cases. But speed is only effective if it can be delivered at massive scale. What this means to enterprises is that they now have a future proof option to replace legacy infrastructures that were not built for today’s modern applications, data sets, or even storage media.

How do you protect data? And what about replication between array and to the cloud?
Vexata provides data protection through snapshots and clones that provide space efficient copies that can be used for data protection and data repurposing. Vexata also integrates with application level replication from SQL Server, Oracle and other primary applications. In mid-2018, we will introduce native array replication that supports sync and async replication across remote sites for disaster recovery.

Who is your competition? Who did you beat recently and which vendors you have replaced?
Our biggest competition is really finding where the application needs are the greatest. Even though we’re a young company, we have been able to prove our technology in some very harsh customer environments. We have excellent success when we engage with the right infrastructure and application architects, because they are trying to solve some very difficult problems. We have successful taken out a number of the early all-flash arrays at several large Fortune 1000 companies.

What is your plan regarding a cloud approach?
Cloud is an important part of our strategy and Vexata is supporting customers with private cloud deployments today, but leverage a cloud based analytics system that is fully integrated with the VX-Manager platform.

What are the use cases you address? I read that you can do both OLAP and OLTP, what you name HTAP (for Hybrid Transaction and Analytics Processing), at the same time? How do you do that?
When you look at the requirements of the different workloads, transaction processing requires very fast I/O access and analytics and machine learning requires extremely high bandwidth for large data ingest rates. Vexata’s architecture is unique in the fact that supports transactional workloads such as OLTP, plus heavy throughout for OLAP analytics workloads. This combined capability provides customers with a real time capability to support Hybrid Transactional and Analytical processing workloads (HTAP) simultaneously and at scale. The performance is 10x better than today’s leading AFAs. We support all major workloads such as KX for time series, SAS analytics, Oracle databases and Microsoft SQL data warehouses.

You do a lot of promotion for a file storage solution with IBM Spectrum Scale? Any details and references?
Last September, we announced the Vexata File Solution and the enterprise-harded IBM Spectrum Scale file system software running on up to four standard dual-socket Intel Xeon servers. It’s a powerful combination that supports industry-standard NFSv3 and GPFS protocols at 1 million+ random file IO/s, 50GB/s read and 20GB/s write bandwidth, and up to 180TB of protected capacity. Our partnership with IBM addresses data-intensive applications operating across large unstructured data sets, including financial trading systems, electronic design automation, 3D rendering and image processing, and genomics and drug discovery. We’ll continue to work with IBM on both the marketing and joint-development fronts in the coming years.

Any plan to offer other interfaces built-in in the array such a file or object (I mean S3) interface? Do you get demands for that?
We are seeing some customer request for us to support unstructured data and will continue to expand out portfolio to support these workloads.

How do you price the product?
Because Vexata is able to deploy using commercially available NVMe SSDs, we’re able to take advantage of the market volume and price our system very competitively with others in the market, being able to consistently hit price points between $2-$3 per gigabyte of effective capacity.

How do you sell your product? Channel partners? OEMs?
We sell through a network of top-tier channel partners and work very closely with those partners to deliver complete solutions.

How did you finish 2017? What’s in store for 2018?
Last year was incredible. It was marked by launching the company after 4 years in stealth and the market validation we received was well above our expectations. In the fall, we announced our full product line and product benchmarks that prove that our hardware and software is the new standard of enterprise price/performance. We also signed our first customers and partnerships. In 2018, expect to hear a lot more from us. We’re accelerating our growth on all fronts. We have ambitious international go-to-market plans and we’ll announce several new products, partnerships and customer deals.

Read also:
$54 Million in Funding for Vexata
Total at more than $100 million
2017.09.21 | Press Release
What We Got on Stealthy Start-Up Vexata
In SSD real time array
by Jean Jacques Maleval | 2017.04.14 | News