Making Sense of Alphabet Soup: How On-Premise as a Service Actually Works

Yesterday we launched the on-premises version (called OPaaS) of our venerable storage platform to acclaim – The Register  and a bit of confusion – Forbes.

That confusion is our fault. We got so excited explaining what we’re delivering (it’s awesome) we forgot to tell anybody HOW we’re doing it. Let’s fix that right now. If you ordered the smallest OPaaS deployment we offer today we would ship to you –

 

— 2 48 port 10GB switches
— 1 48 port 1GB switch
— 1 firewall — 2 x86 servers
— A mix of SSD, SATA and SAS drives, 72 in total
— Cables, instructions, miscellany

We have a default mix of drives but if you wanted a specific set then you would get it. We call that entire collection of hardware a ‘Zadara Storage Cloud’. If you work for Widgets Inc. we would call it ‘Widgets Inc.’s Zadara Storage Cloud’. Each x86 server is a ‘Storage Node,’ a Zadara Storage Cloud is a collection of Storage Nodes and the related switches, cables, etc. Every Storage Node in a Zadara Storage Cloud has to be close (in milliseconds) to every other node so a Zadara Storage Cloud is a geographically limited construct. As an example if Widgets Inc. wanted Zadara storage in their San Jose and New Jersey data centers we would deliver two clouds, ‘Widgets Inc.’s Zadara Storage Cloud, San Jose’ and ‘Widgets Inc.’s Zadara Storage Cloud, New Jersey.’

A bit of history – The first Zadara Storage Cloud launched over three years ago at an Equinix facility in Northern Virginia, right next to AWS US-EAST-1, we call it ‘Amazon Web Services East 1’ (so imaginative!). Today it’s over a thousand drives serving NFS, CIFS and iSCSI to nearly a hundred AWS customers. We currently operate more than 15 other publicly available clouds globally including storage for Azure, CloudSigma, other AWS regions, KVH, CoreSite, Equinix and others. We also operate several private Zadara Storage Clouds for our OPaaS beta customers. 

The Zadara software is where the magic is. We’ll get into the details on this blog next week but basically Zadara’s software lets customers build RAID Groups using drives across any number of Storage Nodes. From those RAID Groups customers create block or file volumes and export those to their servers. There is a bit more to it, details to follow.

Every Zadara Storage Cloud, public or private is monitored 24/7 by our operations and engineering teams in Irvine, CA and Yokneam, Israel. We monitor for faults, ship replacement hardware and perform remote software upgrades from these facilities. We also monitor utilization so we can implement a feature that I love, Proactive Delivery. When any Zadara Cloud becomes 70% full we immediately and automatically ship another pair of storage nodes (or as many as are required based on usage trending). If you don’t use them you don’t pay for them. If you only use three drives from the new nodes you only pay for three drives. On-demand pricing for on-premises storage.

A Zadara Storage Cloud can grow to hundreds of Storage Nodes and hundreds of petabytes of storage, and any Zadara Storage Cloud can replicate to any other Zadara Storage Cloud on the planet.

I hope this helps clear up some of the confusion we caused. Questions? Email me at craig@zadarastorage.com. Ready for some storage done right? Get in touch here – https://www.zadarastorage.com/get-started.

Share This Post

More To Explore