NGX Storage ready to shake European storage positions

The recent 47th edition of The IT Press Tour was the perfect opportunity to meet NGX Storage, its CEO Beyhan Caliskan and product management leader Pradeep Ganesan. For most of us, we discovered the company launched in 2015 in Ankara, Turkey, with R&D in Bangalore, India and Vancouver, Canada, with 200 customers in Europe. The business is done exclusively with channels and the team is actively looking for new partners as they prepare something new with a plan to unveil it in February.

In the meantime, I wish to dig a bit in the current product line, that seems to be a NetApp killer, at least this is their target with already many replacements successes.

The firm has designed and developed essentially 2 models of unified storage, the NGX-H and the NGX-AFA.

The NGX-H for Hybrid embeds SSDs and HDDs for a total capacity of 20PB with SAS, SATA or NVMe internal storage connectivity. It is a unified storage system with dual controllers working in active/active mode exposing block with FC and iSCSI, file with NFS and SMB and object with S3. The team has selected Western Digital JBOD chassis connected to 2 access servers. The system runs FreeBSD and uses the NFS kernel module, for SMB it is Samba and for S3, they use MinIO. External network interfaces use 32Gb/s FC HBAs and 100Gb/s Ethernet NICs.

The NGX-AFA develops similar characteristics except it’s a full flash solution with SAS and NVMe connectivity.

On the data service side, the company has made clear choice and the solution is very comprehensive with:

• Inline data reduction with both compression and deduplication with a zero detection mechanism,
• Intelligent tiering at multiple levels in the system and all data are always written first to the flash layer,
• Accumulation write operations for SSDs what they call Random Flash Sequential Disk (RFSD),
• Data Protection with RAID (level 1, 5, 6, 10, 50 and 60) plus triple parity and 4-way mirrors, snapshot, clone and replication over TCP but erasure coding is not yet available,
• Non disruptive upgrade both for the software and the hardware,
• And also encryption with AES-256, WORM and TOE.
  
  

To boost I/O operations the engineering team implements wide-striping to maximize the bandwidth and delivers expected throughput.

In terms of deployment, classic local racks exist but also metro cluster with what they call MetroScale Cluster leveraging 4-way mirroring, 2 on each site.

We discovered a very compelling solution, very rich, and we understand the adoption in their region. Now they’re working on a new generation of block storage, we were super happy that Beyhan and Pradeep broke the news for us for the very first time and a few weeks before the product will hit the news and the market. I will write about it in a few days, it deserves a dedicated article.