Attending my first Cloud Field Day (CFD#4) in San Jose I had a chance to learn of an innovative storage integration product from SoftNAS. I will blog about it in 2 posts, this is the first one.
The introduction was led by the CEO, Rick Braddy, that gave us a company overview.
He had a long experience on Cytrix, noticing that virtual desktop running Windows needed a very large number of IOPS to feed the system. This was (is) true also in a VMware environment, but, as you can imagine, this kind of storage is quite expensive
The concept on which SoftNAS is based is that “storage is not just a bunch of disks”.
The business model is subscription so no upfront needed subscription based.
Data are no more datacenter-centric, but moving across several DC around the globe, on SaaS app, multiple cloud providers not necessarily public. Manage and control these data is hard and critical: “FABRIC for BUSINESS DATA” is the paradigm.
SoftNAS Cloud is a “Data control and Management Platform”, it means a pre-integrated set of capabilities and tools to manage a very large amount of data natively on the cloud, no more locally. This set is a dedicated Data on-Ramp to the cloud (ideally, a full cloud, more realistically a hybrid solution). I’d like to underline that it’s based on an open source project.
The product will take care of 4 kind of problems:
SoftNAS Cloud integrates the following storage sources using the technology shown in the ring:
This image is full of technology, really, integrating all of them has been an impressive work, especially if paired with an easiness of management, kind of drag&drop Java-driven, and this is one of the open source project.
A summary of what previously shown, in a layered way:
So, the core is an open architecture, SoftNAS leverages this architecture automating and integrating all the single components. APIs are part of this process.
Twitter, http, https, hadooop, webservices, redshift, s3, azure, the list is long, could contribute to this amount of data to manage.
After Rick, John-Marc Clark took the stage, illustrating in a deeper detail the product.
SoftNAS Cloud product is based on the Open ZFS file system (a good post about it from Stephen Foskett here), an open source project, and on Apache NiFi, another open source project, a system built to automate flow of data between platforms running in a Java environment.
First goal: avoid cloud vendor locking, giving the user a freedom in choose the best kind of storage at the price and performances he needs, including any on-premises storage.
A lot of problems that companies encounter approaching the cloud are file-system related, mainly for performance degradations and cost to solve it. IoT and machine learning add another level of complexity when moving these data to the cloud – this is much more a Data as a service rather than a Storage as a service. As Joep stated during his presentation, that’s more a consulting solution rather than a technology solution, helping companies moving their workload (not just storage, but more generally data) to the cloud.
One of the goals that Ultrafast supply is to let this migration from and to the cloud, from and to on-premises systems, faster. Same across clouds.
Another primary goal accomplished by the product is make an object storage performant nearly as a block one.
To summarize, customers encountered different problems moving to the cloud, to be solved, by SoftNAS point of view, through 3 different editions of the product: Essential, Enterprise and Platinum. Details are shown below:
In very few words, the essentials edition is dedicated to the companies that need to move to the cloud backups, DR, object storage and so on. Enterprise is for customer that chose to move their main data to the cloud, the production ones with no recoding of their apps. Last, Platinum, is for all data, a complete integration and automation of the company premises with the cloud in the perspective of moving these data across cloud and on-premises – a complete hybrid solution.
We could consider this product as a Private Cloud NAS.
(Second and last part coming soon).