Nimble’s recent announcement that they’ll ship all-flash arrays alongside their hybrid arrays has got me thinking: I’m excited. Nimble’s announcement further validates that organizations need all-flash performance for certain workloads, regardless of the claims that all-flash service levels in a read cache provide performance that is “similar” to an all-flash array. Obviously this isn’t the case any longer.
|All-Flash Array Facts|
Nimble has entered a highly competitive market space. Trust us, we know. As EMC, Dell, Tegile, Pure Storage, Tintri, NetApp / SolidFire and others have already established an all-flash beachhead, it’s essential to offer a differentiated product and draw a clear distinction between Tegile and other all-flash arrays on the market.
Architectural flexibility, highest resiliency and best economics
Did you know that Tegile’s all-flash portfolio offers unmatched architectural flexibility, the highest resiliency and the best economics? The broad portfolio from Tegile ranges from a cost-effective 2U all-flash array with 18TB of effective capacity (T3500) all the way to a high-density system that can accommodate over 1.4PB of effective flash in just 5U.
While Nimble and other vendors have been adding pieces to complete their part of the puzzle, Tegile’s been offering a broad flash foundation with a complete suite of data services for years. Let’s take a deeper look into what makes Tegile the perfect fit for customers who wish to accelerate their transition to an all-flash datacenter.
Tegile is focused on providing customers with three attributes in their storage:
- IntelliFlash OS
IntelliFlash is a log-structured filesystem designed for flash architectures. Data is written into memory and is inline deduplicated and compressed on both our all-flash and hybrid platforms. These techniques not only reduce storage footprint, they also act as a performance multiplier by maximizing the amount of data kept in flash.
IntelliFlash also coalesces blocks of data in memory, which sequentializes writes into transaction groups. These groups are laid down in dynamic block sizes and written to persistent media to match the block size of the SSD and HDD to avoid block misalignment, for example, which is a major performance drain if architectured incorrectly in the OS. This process optimizes the I/O flow, reduces flash media wear and increases write endurance to further improve writes and reads (to avoid a major performance drain – such as that in NetApp’s WAFL).
- Media Adaptable Platform
Our storage platforms have always been designed to take emerging storage media into consideration. Tegile’s IntelliFlash OS is a flexible architecture that future-proofs the customers’ investment by offering faster and more cost-effective media types as they become readily available. Think of it as keeping your exoskeleton, but updating your endoskeleton. Wouldn’t that be nice? It’s how we’re also able to leverage rapid innovations in hardware. For example, we use 8TB flash modules in our IntelliFlash HD product line.
- Patented Metadata Acceleration
Some storage systems combine metadata with data. Over time, with data being modified, deleted, and rewritten, metadata becomes very fragmented. Additionally, certain storage system features, such as deduplication, can cause metadata to multiply and grow rapidly, especially in smaller block (4KB – 8KB) virtualized environments, not to mention bog down performance from the extra cycles required.
Since deploying our first array in 2012, we’ve used a patented process that automatically separates metadata from data and writes are always prioritized in concert with garbage collection. The metadata is then organized, aggregated, and placed on dedicated logical devices using very low-latency storage (DRAM and fast flash) with optimized retrieval paths. This process accelerates advanced data services, such as inline deduplication and compression, pointer-based snapshots, read/write clones, and thin provisioning.
- Multi-Layered Architecture
Tegile combines multiple storage layers to provide unmatched flexibility. The ability to combine a performance layer (DRAM and fast flash) for deterministic latency and high IOPS with an economical dense flash and capacity layer provides scalability to support a wide range of enterprise workloads and in real time. This provides the best $/transaction for customers who wish to accelerate application service levels when performance and capacity requirements are never known ahead of time.
All-Flash AND Hybrid. Application requirements should dictate the changing landscape of your storage architecture. Tegile customers have the flexibility of choosing all-flash, hybrid, and a combination of both — even in the same array, if desired.
For example, a customer who starts with an all-flash array can independently scale for both performance AND capacity—non-disruptively with both all-flash and hybrid flash (flash and HDD) enclosures. This provides customers who may not be able to predict future storage requirements to non-disruptively add more metadata, cache, and data capacity to their existing platforms. This level of flexibility provides unexpectedly low total cost of ownership to our customers. Ask one of them, they’ll tell you for themselves.
- Active/Active Architecture
Tegile all-flash and hybrid arrays have always been built on an active/active controller architecture that allows both controllers to ingest and serve data simultaneously. Not only are you maximizing your investment on day one, you can also create all-flash and hybrid storage pools and isolate workloads to optimize performance and avoid application I/O contention that happens in single storage pool configurations (active/passive). With availability always in mind, pools can be failed-over transparently, with no disruption in data access using storage pool VIPs and multi-pathing techniques.
- End-to-End Data Integrity
IntelliFlash provides true end-to-end data integrity protection by automatically detecting silent data errors or corruptions — and fixes them on the fly. Every block that’s ingested into a Tegile array receives a unique checksum (think of it as a fingerprint).
Let’s say there’s bad data on a drive in the array, when the application requests the data, IntelliFlash compares each block against its corresponding checksum stored in our metadata pool in our performance layer (that’s where our dedupe table’s stored, too). If the block in question does not match the checksum, it then finds a good block based on the metadata pointers and via parity builds the data and returns the good data to the application. This all happens with no user intervention.
End-to-end data integrity is essential because the number one requirement of any storage array is to provide the same, exact data that was stored originally whenever requested.
Since our first product release, Tegile has given customers the ability to scale independently and non-disruptively for both performance and capacity.
This approach is the same regardless of the route the customer takes: hybrid, all-flash, or high density. This flexibility is taken one step further by providing our customers with unmatched scale. A customer can start with an all-flash array and scale for performance or capacity — or both.
This flexibility provides peace of mind because most people may not know what their performance or capacity requirements might be in the future for a given application. One example of this is our IntelliFlash HD platform that provides a performance layer and a dense flash layer. With a 5x data reduction, that’s over 1.4PB of effective flash capacity in a minimal 5U footprint.
- All-Flash to Hybrid Replication
Tegile was the first next-gen storage company to introduce the ability to replicate from an all-flash array to a hybrid array using native array-based replication, reducing the overall cost of business continuity and disaster recovery for any organization. And we’re a fan of Nimble’s newly introduced ability to do the same.
Tegile has always provided customers with pointer-based snapshots where we only store the changes since the last time a snapshot was taken. This saves significant space for customers who take many snapshots to ensure local recovery, which is common with many of our ex-NetApp customers.
Tegile provides read/write clones in which a customer can make an instant copy of a DB for test/dev or cloning for VDI workloads. One major distinction between us and our competition is that Tegile can add hybrid enclosures (flash and disk) to an all-flash array at the primary site, which essentially provides the same economics that you’d get from a hybrid array, but done locally.
As a customer, you have the benefit of storing snapshots and clones locally for near-instant recovery for your mission-critical applications without relying on network bandwidth to retrieve snapshots. You also have the flexibility to send those snapshots to your secondary site where the cost to store them is even less. While we can genuinely appreciate Nimble introducing this feature, we have hundreds of Tegile customers who have been taking advantage of our replication feature for years.
- End-to-End Deduplication
Tegile provides deduplication for both sites (primary and secondary) to ensure that the data being replicated from our all-flash arrays are also deduplicated on hybrid arrays that may be used as a DR repository. So if you’re achieving a 5x data reduction on the primary site, rest assured that with Tegile you’ll also receive that data reduction on the secondary site.
In summary, with Tegile it’s all about AND. We love that Nimble’s caught up with an AFA offering. Tegile has been providing a complete datacenter solution to customers for years and we’ll continue to do so. But, don’t take my word for it; schedule a 1:1 demo with one of our storage experts and see for yourself.