SanDisk Membrain Software User manual

Synopsis
Membrain is mission-critical cache. Increase the density of cache to power server consolidation,
reduce power consumption and lower overall total cost of ownership (TCO). Boost response
times for critical applications with in-memory like performance. Learn more in this white paper.
Membrain Software
Enterprise cache. Built for flash.
Delivers breakthrough
10x server consoli dation
WHITE PAPER

TABLE OF CONTENTS
Introduction ..............................................................................................................................................1
Memcached Overview .........................................................................................................................2
Membrain Architecture .......................................................................................................................2
Hardware Platform ...........................................................................................................................2
The SanDisk Operating Environment (SOE) ..........................................................................2
Memcached Client/Server Protocol ...........................................................................................3
SanDisk Data Fabric .........................................................................................................................3
Flash Management ...........................................................................................................................3
Flash and Network Management ................................................................................................4
Administration .........................................................................................................................................4
Key Capabilities ......................................................................................................................................4
Using Flash Memory to Increase Capacity, Reduce Power, and Provide Persistence .......4
Using Replication and Automated Failover for HA and Disaster Recovery ...............6
Using Backup and Restore to Protect Against Data Corruption or Loss ....................7
Using Containers to Manage Consolidation ...........................................................................8
Using Cloning to Rapidly Deploy and Warm Identical Membrain Nodes ...................8
Using Administrative Services to Deploy, Management, and Scale ..............................9
Network Configuration ...................................................................................................................9
Enhanced Memcached Statistics ..............................................................................................10
Putting It All Together: Membrain Use Cases ..........................................................................10
Reducing TCO through Consolidation ....................................................................................10
Industry-Leading Data and Service Availability ........................................................................ 11
Summary ................................................................................................................................................. 12
Table of Contents

PAGE 1
Introduction
With the explosive growth of high-volume web sites, software-as-a-service (SaaS), cloud
computing, and other emerging web-based applications, datacenter workloads have increased
exponentially. The business opportunities that are created from the deployment of these
new applications are substantial, but the demands they place on the datacenter are daunting.
Challenges include unprecedented data growth, large increases in rack space and power con-
sumption, too much time-sensitive data to process in a rapid and consistent manner, inability
to effectively scale to meet demand, underutilized resources, and severe space,
power, and budget constraints.
A software cache and / or persistent key-value store is now a key part of the software infra-
structure of many datacenters. Unfortunately, the traditional caching solutions (such as the
widely used Memcached protocol) and key-value store products are loosely coupled and lack
overall optimization. These products do not adequately address the performance, capacity,
scaling, availability, and power challenges of supporting new Web applications effectively.
Huge advances in multi-core processors, flash memory, and high-performance networking
offer huge improvements in performance and power at the component level. But these new
component technologies are severely underutilized when datacenters integrate them at the
data-access tier. Serious scaling, performance, power, networking, space, and complexity
issues still remain.
To address these challenges and opportunities, new system architectures and solutions that
match the workload, operating environments, and hardware are needed. SanDisk has taken
a breakthrough, holistic approach to harness these powerful technology advances to create
Membrain: a breakthrough software cache and persistent NoSQL key-value store optimized
to take full advantage of flash memory, multi-core processors, and high-performance
interconnects.
Membrain is a high-performance, highly available, cost-effective, and scalable solution that
meets today’s demanding data-access needs. Relative to traditional Memcached and the other
NoSQL key-value stores, Membrain delivers order-of-magnitude improvements in performance,
scaling, downtime, power, and space, while cutting total cost of ownership (TCO) by more than
half. The net result is more efficient and highly-available datacenter operations and the ability
to create new revenue-producing business opportunities based on rapid access to terabyte-
scale data.
Membrain powers production workloads in leading Web facing, SaaS, cloud, and enterprise
datacenters today. Membrain offers, among other things, the following unique features:
 Industry-leading throughput, capacity, and data and service availability by tightly integrating
the Memcached protocol with the innovative SanDisk Operating Environment (SOE) and
standard server and flash memory, significantly reducing capital expense, power consump-
tion, rack space, and networking requirements;
 Comprehensive data and service availability through a rich set of high-availability/disaster-
recovery (HA/DR) features, including Persistent Caching, Active/Active Replication, RAID,
and Backup & Restore; and
 100% memcapable compatibility with all existing Memcached client libraries, applications,
and tools, making it easy to integrate into existing environments.
SanDisk customers highly value the consolidation and TCO savings afforded by Membrain. For
example, using Membrain instead of traditional DRAM-only caching brought a social networking
company a 10:1 reduction in the number of servers and a media company a 12:1 reduction in
the number of servers needed to service the same workload. Membrain customers enjoy large
savings in TCO, power, space, and application complexity.

PAGE 2
The industry-leading data and service availability provided by Membrain is a key benefit valued
by its users. Some SanDisk customers use Membrain as their sole data repository, without a
database as the primary data store. For these customers, the comprehensive data and service
availability of Membrain is critical. For example, a SanDisk telecom customer stores all the
address books for users of mobile and cable services in Membrain servers. They configure the
Membrain nodes with transparent replication and automatic failover, and they use the Membrain
Backup/Restore facility for periodic backups for recovery from disasters and less catastrophic
forms of data corruption, such as those caused by client application defects.
The rest of this white paper describes: Memcached; the architecture of Membrain; the key capa-
bilities of Membrain; and use cases for Membrain. We conclude with a summary of key points.
Memcached Overview
Memcached is a distributed key-value caching is widely used in high-traffic Web sites.
Memcached is generic, providing caching for arbitrary chunks of data such as results from
database calls or page renderings. Memcached is proliferating because of its effectiveness at
off-loading database and application servers, with dramatic improvement in Web site response
time and throughput.
Memcached is based on a standard client/server protocol which is maintained by the
Memcached open-source community, which also maintains various client libraries, a reference
server implementation, and management, administrative and compliance tools.
Membrain Architecture
Membrain seamlessly integrates the Memcached protocol with the SanDisk Operating
Environment (SOE) (described later in this white paper) and the underlying hardware platform
to provide superior enterprise caching and key-value store performance, capabilities, and
features.
Membrain is based on a system architecture that holistically manages state-of-the-art, enter-
prise-class flash memory, multi-core processors, DRAM, and high-performance networking to
optimize the performance, availability, and TCO of the Memcached client/server protocol. This
architecture was created based on extensive workload characterization, system modeling, and
optimization. The result is a highly optimized cache/key value store that effectively harnesses
these powerful technologies to provide a solution that is scalable, smart, cost-effective, and
green. Web facing datacenters can now leverage the advanced technology of SanDisk’s tightly
integrated solution to quickly and easily manage data growth, decrease business complexity,
and cut TCO.
Hardware Platform
Membrain runs standard x86 servers with flash memory. The recommended hardware platform is:
 Dell, HP, or IBM x86 server or blade with at least 64GB of DRAM
 PCI-e enterprise flash memory or at least 5 SATA/SAS SSDs
 A single 10Gb or multiple 1-Gb Ethernet ports.
The SanDisk Operating Environment (SOE)
The SOE unlocks the full potential of the hardware by optimizing the utilization of all the
available physical resources across a broad spectrum of workloads. It enables full utilization
and linear scalability of multi-core processors, intelligently manages caching from DRAM to
flash, delivers highly parallel read- and write-optimized flash memory access, and manages high
performance replication of cached data between Membrain servers.
The SOE consists of the Memcached client/server protocol layer, SanDisk data fabric, the flash
management subsystem, flash and network management, and administration, as shown in
Figure 1 on the following page.

PAGE 3
Figure 1: SanDisk Operating Environment (SOE)
Memcached Client/Server Protocol
The SOE protocol layer accepts Memcached commands and is a fully Memcapable 100%-
compatible implementation of the standard Memcached client/server protocol. Membrain is
therefore 100% compatible with all client applications. Using the familiar Memcached APIs,
client applications can perform both distributed caching operations and persistent NoSQL
key-value store operations.
SanDisk Data Fabric
The SanDisk data fabric layer implements the fundamental algorithms necessary to maximize
system balance, performance, power efficiency, and availability. The SanDisk data fabric func-
tions include:
 Fine-grained, highly concurrent and efficient multithreading and core management required
to parallelize hundreds of thousands of simultaneous flash and network accesses per second;
 Intelligent, dynamic DRAM caching required for fast and efficient staging of both small and
large objects;
 Creation and management of dynamic containers providing fine-grained control over cached
data, supporting attributes of size, persistence, store vs. cache mode, and replication;
 Transparent, highly optimized data replication and failover.
Flash Management
The flash management layer provides space and persistence management across the parallel
flash devices. The flash management subsystem functions include:
 Mapping object keys to flash locations, employing algorithms that minimize the amount
of mapping data that is kept in DRAM so that most of the DRAM can be used for caching
Memcached and key-value store data;
 Maintaining persistent metadata, enabling the preservation of Memcached data across
system failures; and
 Optimizing data placement and replacement across the parallel flash devices to balance
accesses, maximize effective space usage, and minimize flash wear.
SanDisk Operating Environment (SOE)
Administration
Configure •Monitor •Control •Optimize
Memcached client/server protocol
100% compatible
Data Fabric
Thread & Core Management
Synchronization/Concurrency Management
DRAM Cache Management
Container Management
Replication Management
Flash Management Subsystem
Space Allocation
Object Replacement (Cache Mode)
Persistency Management
Flash & Network Management
Highly Concurrent I/O & Messaging
Data Striping, RAID

PAGE 4
Flash and Network Management
The flash and network management layer provides highly parallel, reliable flash-access and
network-messaging facilities, optimized to exploit the full I/O capacity of the flash and
networking hardware for high performance and highly available caching services. The flash-
and network-management layer functions include:
 Highly concurrent, low latency network and flash access;
 RAID 5 across parallel flash devices, enabling device failure tolerance and hot swap while
minimizing capacity overhead for redundancy; and
 High-performance, inter-server messaging enabling failure tolerance and live upgrade
through synchronous replication.
Administration
The administration services make Membrain easy to deploy, manage, and scale. The administra-
tion layer functions include:
 Full-featured graphical user interface for easier management and monitoring of the
Membrain instances
 Support for monitoring instances of stock memcached, which are often used in conjunction
with flash-optimized Membrain
 Compatibility with existing tools and applications;
 Hot-key monitoring providing visibility into frequently used keys and highly active clients;
 Full and incremental backup and restore of cached data; and
 Remote presence, predictive failure analysis, and automatic restart.
Key Capabilities
Using Flash Memory to Increase Capacity, Reduce Power, and Provide Persistence
Membrain integrates DRAM and flash to meet the throughput and scaling requirements of
Memcached and key-value store deployments at much lower cost than a DRAM-only solution,
in terms of both capital expense and power and space consumption. SanDisk’s flash-memory
subsystem also enables the following new capabilities with Memcached:
 Order-of-magnitude increase in capacity per server;
 Order-of-magnitude reduction in power consumption per Memcached or key-value store
operation;
 The ability to persist data across shutdowns/power failures.
As shown in Figure 2, SanDisk’s highly parallel, optimized flash-memory subsystem increases
Memcached capacity by an order of magnitude over a traditional Memcached server. This is
especially significant because many Memcached installations are capacity limited. For example,
an application may require a total of 1 TB of memory to achieve a targeted hit rate. If a typical
Memcached server has 64 GB of DRAM, this application would require 16 nodes to get the
required capacity. Spreading the client workload across 16 nodes results in poor processor and
network utilization (<10%). Flash memory allows the entire working set of the application to be
hosted on a single node, enabling balanced use of processor and networking resources through
the SOE.

PAGE 5
Figure 2: SanDisk Cache/Store Capacity Increase through Flash Subsystem
Figure 3 shows the throughput of Membrain relative to Membase in single node and replicated
2 node configurations. The test was performed with dual-socket Westmere servers with 1/2 TB
of flash memory. The workload has an average object size of 2kB executing 90% single gets
and 10% puts. Membrain achieves over 5 times the throughput in single node and replicated
2 node configurations.
Figure 3: Throughput Comparisons of Membrain vs Membase
Membrain delivers large consolidation relative to alternative NoSQL key-value stores, since
their performance drops dramatically when data does not fit in the server DRAM. This is shown
in Figure 4 which compares the single-node performance of Membrain, SanDiskSQL™ (a full
high-availability build of MySQL and its InnoDB storage engine), Cassandra, and MongoDB.
Each is executing the same key-value random query benchmark of 32 million and 64 million
1 KB items, on the same hardware (dual quad-core Intel processors with 64 GB of DRAM and
512GB of flash memory). The in-memory key-value random query performance of Membrain
greatly exceeds other NoSQL alternatives, giving from 2 to 100 times improvement. The per-
formance of Membrain when data cannot fit in DRAM is at least 25 times that of the alternative
key-value stores.
Figure 4: Throughput Comparisons of Key Value Store Alternatives
Cache capacity per node (GB)
100
200
0
300
400
500
600
SanDiskTraditional Memcached
50,000
100,000
0
150,000
200,000
250,000
300,000
Membase
1.6.0.1
SanDisk
Membrain 3.0
Membase
1.6.0.1
SanDisk
Membrain 3.0
Single Node
Persistent Key Value Store
Replicated
Key Value Store
in DRAM in Flash
Cassandra 10,500 1,790
MongoDB 49,000 4,000
Membrain 310,000 160,000

PAGE 6
This high consolidation significantly reduces the cost, power, cooling, and space required for
Memcached or key-value store operation. As will be shown in the section “Reducing TCO
through Consolidation”, the power efficiency of flash, combined with a balanced use of all system
resources, results in an order-of-magnitude improvement in Memcached throughput per watt.
Since flash memory retains data when powered down, it allows Memcached and key-value store
data to be preserved across system shutdowns or power failures. This can dramatically reduce
the time it takes to re-populate Memcached after a restart, reducing node downtime. Clients
can initiate Sync operations to Membrain to explicitly control the persistent consistency state.
Using Replication and Automated Failover for HA and Disaster Recovery
In many applications, Memcached and key-value store availability is crucial to delivering the
quality of service required by Web sites. If a Memcached server fails, the application must
go to the primary data service, such as a database, to retrieve data. Database accesses are
much slower than Memcached accesses, a problem exacerbated by the extra load when a
Memcached server is down. Therefore Memcached outages can cause severe degradation in
application throughput and response time. With key-value stores, the store often represents
the primary source of the data. So in the event of a failure, the service will be interrupted.
Membrain solves the problem of availability in the event of planned or unplanned downtime.
Each Membrain server is assigned distinct virtual IP addresses (VIPs). Two Membrain servers
can be configured as a mirrored pair, in which data stores (Memcached set/put/replace, etc.)
to one server are replicated to the other. If one server in the pair goes down, the other trans-
parently assumes its VIPs and serves its clients.
As discussed in more detail later in this white paper, Membrain supports the creation of multiple
virtual storage domains called “containers” to provide fine-grained control over cached resourc-
es. Containers provide isolation and policy control, yet efficiently share processor, DRAM, and
flash resources. Containers can be configured in eviction or store mode as well as persistent or
non-persistent mode. Membrain replication supports all container modes.
Figure 5 shows how Membrain replication works. On each write to a node, the Membrain
messaging service transparently replicates the write to the other node. Replication is done
synchronously so that when a client receives a positive acknowledgment for a write, the data
is guaranteed to reside on both nodes in the pair. The heartbeat function of the Membrain
messaging service detects failures when the messaging connection closes or times out. When
this occurs, the surviving node activates the VIPs of the failed node so that it can take over
servicing requests for the failed node.
When a failed replica comes back into service, it initiates actions to make itself current. First,
it notifies the surviving node that it is ready to recover. The surviving node then starts copying
the necessary data to rebuild the recovering cache, and also starts replication of all new write
operations. When the recovering node has finished repopulating its Memcached and key-value
store data, it takes ownership of its VIPs again and resumes service.
Replication is active-active: writes to a particular container can be sent to either node in a
mirrored pair. Although writes to the same key could be sent to either node, this can result in
inconsistent Memcached caches or key-value stores (i.e., Node 0 has different contents for the
key than Node 1). To maintain consistency, applications should map distinct key sets to each
node in a mirrored pair.
Membrain replication can also be used to perform a “rolling upgrade”, in which the supported
application is updated to a newer version without a service disruption. This is done by taking
one node in a mirrored pair offline, upgrading it, and bringing it back online. When the upgraded
node is back online it goes through the recovery process with the surviving node, and then
resumes service.
 Active/active replication fully utilized all nodes
 Synchronous replication ensures no data is lost during outages
 Automated failover delivers continuous service availability, and eliminates the need for
manual, error prone user intervention
 Transparent to client side apps and requires no additional overhead

PAGE 7
Figure 5: SanDisk Replication
If applicable, the SSDs in a Membrain node can be configured using RAID 5 to further increase
node availability. This allows the node to tolerate the failure of a single SSD without a loss of
service. When using SSDs that are hot-swappable, a defective SSD can be replaced without
stopping the Memcached / NoSQL service.
Using Backup and Restore to Protect Against Data Corruption or Loss
Many applications need to back up the contents of Memcached or key-value stores to recover
from user errors or application-driven data corruption, for full recovery and restore from
catastrophic data loss, for warming the caches of new servers before bringing them online, and
for logging.
Membrain supports full and incremental backup and restore of persistent containers to hard
disk drives. A full backup is a logical copy of all objects in a container. An incremental backup
is a logical copy of objects in a container that are new or have changed since the previous
backup, including a logical representation of deleted objects. A full backup is taken to start a
new backup “series”, which contains the full backup plus zero or more incremental backups.
There is no limit to the number of incremental backups in a series. Backups can be scheduled at
regular intervals.
Backups can be taken while the server is still servicing client requests. In this case, data written
by a client after the backup is started may or may not be included in the backup.
Restoring a backup is the process of replaying backup streams to a server. A backup can be
restored to any container. The target container must already exist and have sufficient space
before the restore is started. The container can be restored to an arbitrary checkpoint within
the series. See Figure 6.
Figure 6: SanDisk Backup and Restore
gets/sets
to key set 0
Master 0
Synchronous Replication
Master 1
VIP group 0 VIP group 1
gets/sets
to key set 1
Normal Operation
Master 0
Re-sync data
Synchronous Replication
Master 1
VIP group 1
gets/sets
to key sets 0
and 1
Recovery
Master 1Service
Interruption
VIP group 0
VIP group 0
VIP group 1
gets/sets
to key set 1
Failover
Container
TCP/UDP Port TCP/UDP Port
Hard Drive Hard Drive
TCP/UDP Port TCP/UDP Port
Cap:
128GB
Non-Persistent
Cache Mode
Data:
Thumbnails
Container
Cap:
64GB
Persistent
Cache Mode
Data:
Queries
Container
Cap:
64GB
Persistent
Store Mode
Data:
Cookies
Container
Cap:
64GB
Persistent
Store Mode
Data:
Photos
Backup
Enable backup
per container
Full or
incremental
backup
Enable up to
8 parallel
streams
Choose
backup interval
Container
TCP/UDP Port TCP/UDP Port TCP/UDP Port TCP/UDP Port
Cap:
128GB
Non-Persistent
Cache Mode
Data:
Thumbnails
Container
Cap:
64GB
Persistent
Cache Mode
Data:
Queries
Container
Cap:
64GB
Persistent
Store Mode
Data:
Cookies
Container
Cap:
64GB
Persistent
Store Mode
Data:
Photos
Restore
Restore to
empty container
Restore to any
previous full or
incremental
backup point

PAGE 8
Using Containers to Manage Consolidation
Membrain supports the creation of multiple virtual storage domains called “containers”, provid-
ing fine-grained control over cached and stored resources. The container paradigm enables:
 Isolation of separate applications or datasets based on usage patterns;
 Policy specification and control for capacity, replacement, persistence, and availability; and
 Dynamic sharing of processor, DRAM, flash, and network resources.
Containers are created and managed through the SOE administrative services, with the
container attributes being dynamically enforced by the SanDisk data fabric. A single Membrain
node hosts multiple configurable containers. The TCP/IP addresses are shared by all containers
within a Membrain node.
Each container has distinct attributes: the port through which it accepts client requests; evic-
tion policy; storage capacity; and persistence. Each container has distinct ports for accepting
Memcached requests so behaves like an independent instance of Memcached. However, since
the processors, DRAM, flash, and networking are all dynamically shared across containers, the
result is a much more balanced and efficient use of resources than with multiple traditional
Memcached instances.
One container attribute is persistence. Using legacy Memcached, volatile DRAM is used to
cache data. Membrain has introduced the ability to make cached data durable. With Persistent
Caching, the cached data is retained in the flash memory and is available even after a power
outage, allowing Membrain to instantly recover to its peak performance, avoiding performance-
degrading, multi-hour or multi-day cache warm-up periods. Another container attribute is Store
Mode. Store Mode allows client-side applications to have control over eviction of data from the
cache. Data will not be evicted without the application initiating it. See Figure 7.
Figure 7: Membrain Node with Multiple Configurable Containers
The SOE, with its advanced processor, DRAM, flash and interconnect management algorithms,
enables access to highly-available terabyte-scale flash memory at DRAM-like speeds. Persistent
Caching in conjunction with Store Mode allows Membrain to be used as a persistent object
store, enabling powerful new applications using the Memcached API and protocol.
Using Cloning to Rapidly Deploy and Warm Identical Membrain Nodes
“Cloning” is the process of copying the contents of an existing Memcached server to a new one
before putting the new one into service. It is an efficient way of warming a Memcached server
in an installation where multiple servers hold the same data. The server that is cloned from can
remain active while the cloning is in progress, with all new writes propagated to the new server.
A clone operation can also be performed as a backup followed by a restore, though it may not
be as efficient.
Container
TCP/
UDP Port
TCP/
UDP Port
TCP/
UDP Port
TCP/
UDP Port
TCP/
UDP Port
Cap:
128GB
Non-Persistent
Cache Mode
Container
Cap:
64GB
Persistent
Cache Mode
Container
Cap:
64GB
Persistent
Store Mode
Container
Cap:
64GB
Non-Persistent
Store Mode
Data:
Photos
Container
Cap:
64GB
Non-Persistent
Store Mode
Data:
Session ID
Container

PAGE 9
Using Administration Services to Deploy, Manage, and Scale
Open-source tools for simplifying the administration of Memcached are very limited. That is
why Membrain includes an easy-to-use graphical user interface (GUI) that allows all aspects
of Memcached/NoSQL cluster configuration and maintenance to be performed with intuitive
screens and menus, including:
 Network configuration;
 Adding and deleting Memcached containers;
 Adding and removing nodes to or from the cluster;
 Enabling and disabling replication;
 Restarting and recovering a failed node;
 Saving and restoring cluster configuration data;
 Backing up and restoring Memcached data;
 Querying server statistics in graphical form;
 Starting and stopping Memcached;
 Performing rolling upgrades; and
 Hot-key and hot-client statistics.
Figure 8: Membrain GUI
Membrain also has a comprehensive command line interface (CLI) for users who prefer to
integrate Membrain into an existing administration framework. Membrain is compatible with
standard cluster administration tools, including Ganglia, Nagios, Cacti, and SNMP.
Network Configuration
Memcached installations use a wide variety of network configurations involving various forms
of virtual IP addresses (VIPs), virtual LANs (VLANs), various types of interface bonding, and
redundant link configuration for fault tolerance. Traditional Memcached provides no support
for network configuration; the onus is on users to manually configure the networking on each
server. Membrain provides an easy-to-use interface for configuring multiple Ethernet interfaces,
virtual local area networks, interface bonding (single and multiple), and virtual IP addresses.

PAGE 10
Enhanced Memcached Statistics
Membrain provides visibility into the state of Memcached/NoSQL servers, including statistics on:
 Hot keys;
 Hot clients;
 DRAM and flash space utilization;
 DRAM and flash access and miss rates;
 DRAM and flash eviction rates;
 Processor utilization;
 Network and flash I/O utilization; and
 Read and write response times.
Membrain provides access to Memcached server statistics via server logs and automated
charting utilities in Ganglia and Nagios.
Hot-key and hot-client statistical data provides sorted lists of the most frequently used keys
and the most active clients. This data is crucial for identifying client and cluster configuration
problems. Membrain implements an embedded user-mode analyzer for maximum performance.
Putting It All Together: Membrain Use Cases
This section shows how the extended capabilities of Membrain solve customer problems in
typical Memcached/NoSQL deployments through:
 Reducing TCO Through Consolidation;
 Industry-Leading Data and Service Availability.
Reducing TCO through Consolidation
Figure 8 shows the Consolidation and Reduction in Total Cost of Ownership (TCO) which is
realized in a 2-TB Memcached/NoSQL customer scenario. Consolidation provides a reduction
in the number of servers from 32 to 4, coupled with a reduction in power consumption from 18
kW to 2.5 kW. TCO is reduced by 50% over three years with the consolidated SanDisk solution.
Figure 9: Consolidation and TCO Reduction with Membrain
The TCO comparison includes initial capital expense, maintenance, and the cost of “rack, power,
and pipe” over a three-year interval. The “rack, power, and pipe” estimates use rates from
a typical Web hosting site. The TCO computation only includes tangible costs; it does NOT
include any cost savings due to the simplified administration for an 8:1 server reduction.
SanDisk delivers
50%+
reduction in TCO
3 year TCO for 2TB
memcached workload
OpEx: $610,000
CapEx: 32 servers
OpEx: $282,000
CapEx: 4 SanDisk-powered servers
vs

PAGE 11
This analysis is based on a configuration of 32 traditional Memcached servers with 32 GB of
memory each. It uses a standard client workload with 1-kB average object size (randomized),
128-kB average key size (randomized), 95% gets (with 1 object per get) and 5% sets. Traditional
Memcached servers, limited by DRAM capacity, only use about 1 Gb of network bandwidth and
only use a small fraction of the Nehalem processors. The high utilization and system balance of
the SanDisk architecture enable the consolidation and the power efficiency of SanDisk parallel
flash memory yields an eight-fold improvement in throughput per watt.
Industry-Leading Data and Service Availability
Web sites increasingly depend on Memcached and key-value stores for top application per-
formance. Failures in this tier can lead to large, unexpected loads on the underlying database,
which can significantly degrade the end-user experience. Membrain provides a comprehensive
set of HA/DR features to ensure a consistent user experience through both planned and
unplanned downtime.
Some SanDisk customers use Membrain as their sole data repository, without a database as the
primary data store. Whether or not this is the case, the high availability provided by Membrain
is a key benefit for many of SanDisk’s customers.
Membrain enables a range of availability options for Memcached and key-value store deploy-
ments. As shown in Figure 9, the Membrain availability extension can reduce the downtime
more than 90% relative to a traditional Memcached cluster.
Figure 10: SanDisk Availability Spectrum
Persistent Caching
 All cached data is immediately available, even after a power outage;
 Delivers the persistence of flash memory with the performance of DRAM;
 Avoids performance-degrading, multi-hour or multi-day cache warm-up periods;
 Reduces system downtime by more than 40%.
SSD RAID
 Protects from downtime caused by SSD failures and planned SSD maintenance;
 Minimizes the capacity overhead required for redundancy.
 Reduces system downtime by almost 60%, when combined with Persistent Caching.
Traditional
Memcached
SanDisk availability extensions (Cumulative effect)
Reduction in downtime (%)
Persistent
Caching
SSD
RAID
Replication &
Failover
10
30
20
0
40
50
60
80
100
90
70

Replication & Failover
 Active/active replication fully utilizes all nodes;
 No data is lost during outages, because of synchronous replication;
 Automated failover delivers continuous service availability and eliminates the need for
manual, error-prone user intervention;
 Failover is transparent to client-side applications and requires no additional overhead;
 Reduces system downtime by more than 90% when combined with Persistent Caching and
SSD RAID.
Backup & Restore
 Full and incremental backups can be scheduled to the onboard HDDs;
 Enables the cache data to be restored to a previous state in the event of application-driven
data corruption, user error, or other data loss.
Summary
In conclusion, Membrain delivers an entirely new level of enterprise caching and key-value store
capability relative to traditional Memcached and other NoSQL alternatives. Membrain provides:
 Vastly Higher Per Node Capacity and Throughput: Eight times the capacity and throughput
of traditional Memcached servers;
 Significantly Lower TCO: Server capex and opex reductions cut TCO by more than 50%
before factoring in the value of much less downtime and administration overhead;
 Quick Deployment: Easy plug-and-play installation and configuration;
 Seamless Operation: 100% compatible with existing client applications and management
tools;
 Highest Reliability: Delivers enterprise-class reliability with innovative persistence, replica-
tion, and recovery software;
 Easy Management: Simple, yet powerful centralized management and reporting with
integrated CLI and GUI;
 More Revenue: Powers new revenue-producing applications enabled by fast access to
terabyte-scale data;
 World-Class Support: SanDisk provides 24/7/365, single-point-of-contact, Membrain service
and support.
 Support on popular OS platforms: RHEL 5.4, 5.5, 6.2 and CentOS 5.4, 5.5, 6.2
For more information, visit www.sandisk.com/enterprise
© 2012 SanDisk Corporation. All rights reserved. No part of this document may be reproduced by any means
nor translated to any electronic medium without the written consent of SanDisk Corporation. Specifications
are subject to change without notice. SanDisk is a trademark of SanDisk Corporation, registered in the United
States and other countries. Schooner Information Technology and the Schooner Information Technology logo
are trademarks of SanDisk Enterprise IP LLC. All other names mentioned herein are for reference purposes only
and may be the trademarks of their respective holder(s).
SanDisk Corporation
Corporate Headquarters
601 McCarthy Blvd
Milpitas, CA 95035 USA