High Availability vs. Budget Availability

For many organizations high-availability is non-negotiable, for others it is a battle between high availability vs. budget availability. If resources and budgets were no object, all storage would be replicated synchronously across multiple nodes, then geographically across data centers, and even include replication to the cloud.

If you have mission critical data, StorONE offers and highly recommends all of these High Availability (HA) data protection methods. As we detailed in “Requirements for Extreme High Availability” we deliver the most flexible, resilient and affordable HA in the industry. But if you have locations that need to run in a non-replicated, single node configuration because of tight budgets, or environmental constraints (such as limited power or connectivity)–then StorONE’s DirectWrite provides the safest possible single controller solution on the market today. 

High Availability vs. Budget Availability

Write Caches Require High Availability

One of the primary reasons that most storage systems on the market today are “H/A-Only” is they employ write-caches. If a storage system is acknowledging writes before storing that data on persistent media, then it needs to also mirror that write to another controller in case the first controller fails. There are other reasons for dual controller configurations but for systems that require a write-cache crutch there is far too much data-loss risk to attempt such a solution with a single controller.  

At StorONE, we think that the time of the write-cache or any form of write-acknowledgement prior to writing to persistent media has come to an end, just like All-Flash Arrays. To deliver a cacheless-architecture takes a fresh approach to storage software design, which is why we rewrote storage software from the ground up to use today’s advanced storage media optimally.  The result is the cache-free S1 Storage Engine.  

Legacy storage stacks introduce both latency and unnecessary points of failure via a very common workaround for sub-optimal use of persistent media, called write caching. Increases in CPU (controller) and RAM performance (RAM, MRAM, NVRAM, NVDIMM) allow many vendors to be lazy with their code, and write first to RAM, then stage to MRAM, and ultimately deposit writes upon the persistent media layer. This adds quite a bit of cost, and all the hops and acknowledgements circumvent most of the performance gain they had hoped to achieve. .

More troubling, since the acknowledgements are sent from RAM, additional failure points are created, especially worrisome if the array is in a location where power may be less reliable than in a data center. For example, in an on-premise closet in a regional or remote location. In the event of a power failure or server crash, all the data in the RAM used for the write-cache is lost with no way to recover it. Many mid-sized organizations also can’t afford an enterprise-class data center and they are equally vulnerable to data loss from write-caching. 

The S1 Engine Enables HA Protection, From a SINGLE Storage Node

StorONE spent eight years developing  our unique storage engine, solving many common storage constraints in the process. A key design goal was to offer high data availability, even in the case of a single, StorONE storage node. 

The obsolete legacy storage software was re-written from the ground up with new algorithms, architected leaner, and the old IO stack flattened into a single efficient layer, so we could take direct advantage of the high-performance, and affordable, flash media available today. Using the S1 Engine with modern media such as Optane™, NVMe and SAS SSDs, latency is so low, write-caches are unnecessary. As we discuss in our blog, “Write-Caches Are Unnecessary,” StorONE’s DirectWrite completely removes the need for a write cache! 

When DirectWrite is combined with other powerful features in the S1 Engine, such as our vRAID, and our powerful, embedded S1:SNAP, (which offers unlimited snapshot protection against ransomware or human errors as frequently as every 3 minutes, and effectively replaces costly and slow backup software), suddenly a single node system begins to approach the data integrity and uptime of full blown, multi-node, HA systems running in enterprise data centers. 

High Availability vs. Budget Availability

If you factor in that server quality is at an all time high and the availability of sub 4 hour response times, it makes a single controller option more palatable. Certainly if a single node system fails, you are down for potentially four hours. It is part of the high availability vs. budget availability risk equation. The good news is with StorONE, you can easily and non-disruptively add a second node for full HA the next time budget dollars become available.

If you are facing the high availability vs. budget availability decision then data tiering should also be a key factor in your selection criteria. StorONE offers intelligent, automatic, single node data tiering. It is very cost effective and efficient to DirectWrite to a large flash tier, and then to later sequentialize, and de-stage that data to low cost hard disk drive media. A detailed discussion is available in our latest spy novel “The Hard Drive Conspiracy” StorONE saves you budget dollars up front and continues to deliver savings throughout the solution’s life. We provide the lowest total cost of ownership in the storage industry. To learn more, download our white paper “The TRUE Value of Total Resource Utilization.”


DirectWrite allows a SINGLE NODE to provide the high data-availability of expensive, multi-node competitive systems, because it eliminates write caching and its costly overhead.

The DirectWrite feature is made possible by the StorONE Engine, which is the foundation of StorONE’s S1:Enterprise Storage Platform. It is a crucial feature that delivers maximum write performance while ensuring a very high level of data integrity. With DirectWrite, the application receives the write-acknowledgement from the persistent media, not from a cache that later has to move data. Generating write-acknowledgement from persistent media guarantees the highest level of data integrity. Without first ensuring data integrity, all other advanced data protection capabilities are essentially worthless. 

To learn more watch our on-demand virtual whiteboard session “The Problems with Storage System Write Caching

Posted in
Brett Schechter

Brett Schechter

Brett has 20+ years of storage experience, holding executive, and direct roles in product management and product marketing. Prior to joining StorONE, he has led storage purchasing groups at major cloud providers, giving him insights as a consumer of high end storage systems.

What to Read Next

Better Data Protection and Resiliency

The number one responsibility for any enterprise storage system is to protect your data. Delivering data resiliency is job one for the StorONE Enterprise Storage Platform. The high performance and low-cost of the platform have no value if you can’t have complete data confidence. You can achieve better data protection and resiliency with an Enterprise […]
Read More

The Write Cache Crutch

Most storage systems create a write-cache using system RAM to accelerate performance. The write cache crutch enables these systems to improve performance. Like most crutches, however, it creates dependencies that put data at risk and complicate system design. The motivation for a write cache is simple. Most systems have poor performance when writing directly to […]
Read More

RAID is Not Data Protection

RAID is not data protection. It does not provide adequate protection from multiple media failures. The technology also doesn’t offer rapid recovery, especially as drive sizes increase. In addition, RAID doesn’t adapt well to changes in the environment. RAID doesn’t protect against multiple forms of failure, can’t be counted on for rapid recovery, and adds […]
Read More

Learn More About the Hidden Cost of Dedupe

  • This field is for validation purposes and should be left unchanged.