Enterprise Storage Stack
(ESS) is designed to protect your data. A number
of techniques are used to create high
reliability storage. Additional options are
available to further enhance data protection and
reliability.
Design Fundamentals:
All data written by ESS is written in a FIFO
linear stream to free space on the SSD set. This
means that blocks are never updated in place. In
place block updates are a common cause of data
corruption in many designs. ESS gives you the
benefits of a Copy On Write (COW) architechure
without the overhead. ESS is also inherently
designed to maintain the chronological ordering
of updates. In many cases, chronological updates
have very high overhead. Because of the linear
nature of ESS writes, chronological updates are
easy to maintain at full device bandwidth. This
makes ESS storage inherently "crash clean"
protecting file systems and other data
structures.
Error Detection and
Correction: one optional feature of ESS
is the use of hashes to check data reliability
on read, together with the use of micro-stripes
to allow the repair of damaged data. This free
feature is only available to OEM licensees.
Dual System High
Availability: Dual system High
Availability is intrinsically available to all
customers. In the dual system environment, data
received by one system is handed off to a second
system, and acknowledgement of data receipt to
the client does not occur until the duplicate
system confirms data receipt. Accordingly, data
from a client is stored on two systems rather
than one.
Proper
configuration of a SAN then permits automatic
fail over to the second system in the event the
first system fails. The robustness of the system
pair can be enhanced through the use of Raid-5
or Raid-6 media configuration with almost no
performance loss. Similarly, dual-system HA is
generally less expensive than Dual-Port HA
solutions for any storage environment up to 10TB
of capacity. Beyond this point, Dual System HA
with its multiple disk sets has greater
potential redundancy than can be achieved with
Dual-Ported solutions.
Dual-Port, Dual
Controller High Availability: Dual-Port
solutions allow two controller cards in a single
rack-mount chassis, or alternately two separate
systems with external JBOD connectivity, to
manage one set of SAS SSDs, or a set of SATA SSDs
with interposer cards, or a set of NVMe SSDs
configured with split PCIe ports. This technique is
preferred by some customers and OEM licensees
because it creates full redundancy, most
commonly in a single chassis package. ESS
special modules assure that the memory contents
of one system, including new incoming data, are
replicated on the second controller, and
maintain almost instant switch-over when the
primary system fails. This software is currently
only available to OEM licensees.
Real-Time Logger
Device: While our FIFO method satisfies
the needs of most situations, some people want
the assurance that all data received by a system
is physically committed to hardware. To this
end, ESS supports a a variety of nvRAM hardware
solutions including PCIe memory card with
built-in battery or SuperCap backup power. These
cards only need to protect the small amount of
data "in flight" during writes. This
hardware/software solution is currently only
available to OEM licensees.