Role of Cache Memory in EMC Clariion.

The EMC CLARiiON uses cache in a manner similar to how cache memory is used in traditional workstations and servers. With servers and workstations, application code and application data locality and immediacy are the focus of the design. A characteristic of many applications (including file systems) is to cache new data locally, then periodically flush the data to the actual storage device. This “lazy write” approach would result in large bursts of large I/Os to the storage systems - a perfect fit with the “burst smoothing” benefit of CLARiiON’s caching. Cache page size is perhaps the most influential parameter on cache performance. The division between staging and storage memory is not formally defined by addresses, but more by functionality - as data arrives, it is always staged, and as needed, it may be marked for storage as well. The Partition Memory dialog in Navisphere Manager enables part of the cache to be used for storage operations. Unlike the Symmetrix, certain writes may bypass cache memory.

Salient Features :-

1) Cache memory on an SP performs two tasks.

Staging: Temporary buffering of current read and write data.

Always performed on each I/O.

Storage: Repository for frequently accessed data.

Maintaining copies of read and write data.

User must explicitly enable this (for both read and write).

2) Benefits of caching

Burst Smoothing - Absorb bursts of writes without becoming “disk bound”.

Write cache optimization.

Locality - Merge several writes to the same area into a single operation.

Increases write performance.

Immediacy - Satisfy user requests without going to the disks.

Read cache optimization prefetching of data for sequential reads.

Know Difference Between SAN Versus DAS.

SANs make effective use of Fibre Channel networks and IP networks to solve the distance and connectivity problems associated with traditional DAS solutions such as parallel SCSI. In a SAN, a device can be added or removed without any impact on I/O traffic between hosts that do not participate in the configuration change. A host can reboot or disconnect from the SAN without affecting storage accessibility from other hosts. New arrays can be added to the SAN,and storage from them can be deployed selectively on some hosts only without any impact on other hosts. Thus, SANs enable dynamic, non-disruptive provisioning of storage resources.

SAN architecture allows for multiple servers to easily share access to a single storage array port. This is technically possible with parallel SCSI too, via the use of daisy-chained cables. However, the setup is static, physically cumbersome, subject to practical constraints from requirements on signaling integrity, and difficult to establish and maintain. 

SAN architecture also allows for a single host to easily connect to a storage frame via multiple physical and logical paths. In a multipathed configuration, and with the use of multipathing software such as Powerpath, the host experiences I/O failures only if every one of its logical paths to the storage array fails. Multipathing software can also help balance the host’s I/O load over all available paths. Multipathing capability thus allows for the design of a high performance, highly available, redundant host system.

SANs make it simple to consolidate multiple storage resources – such as disk arrays and tape libraries - within a single physical or logical infrastructure. These resources can be selectively shared across host computers. This approach can greatly simplify storage management, when compared to DAS solutions.

Know EMC FLARE Operating Environment.

FLARE software manages all functions of the CLARiiON storage system. Each storage system ships with a complete copy of FLARE software installed. When you power up the storage system, each SP boots and executes FLARE software.Access Logix software is optional software that runs within the FLARE operating environment on each storage processor (SP). Access Logix provides access control and allows multiple hosts to share the storage system. This “LUN Masking” functionality is implemented using Storage Groups. A Storage Group is one or more LUNs within a storage system that are reserved for one or more hosts and are inaccessible to other hosts. When you power up the storage system, each SP boots and executes its Access Logix software. Navisphere Management software is a suite of tools that allows centralized management of CLARiiON storage systems. Navisphere provides a centralized tool to monitor, configure, and analyze performance. CLARiiON can also be managed as part of EMC ControlCenter, allowing full end-to-end management.

Salient Features:-

1) FLARE Operating Environment runs in the CLARiiON Storage Processor.

– I/O handling, RAID algorithms.

– End-to-end data protection.

– Cache implementation.

2) Access Logix provides LUN masking that allows sharing of storage system.

3) Navisphere middleware provides common interface for managing CLARiiON.

4) CLARiiON optional software including

– Access Logix.

– MirrorView, SnapView, SAN Copy.

5) EMC ControlCenter provides end-to-end management of a CLARiiON.

6) FLARE performs provisioning and resource allocation.

7) Memory budgets for caching and for snap sessions, mirrors, clones, copies.

8) Process Scheduling.

9) Boot Management.

What is a Storage Processor ?

The main component in all CLARiiON series arrays is the Storage Processor. Storage Processors (SPs) are configured in pairs for maximum availability and are Field Replaceable Units (FRUs). SPs provide both front-end connectivity to the hosts and back-end connectivity to the physical disks. Each Storage Processor also includes up to 4 GB of memory, most of which is used for cache. Cache memory is segmented into read cache memory and write cache memory. Read cache memory is used for staging and prefetching read requests from the host. Write cache is used to accelerate host writes to the storage system. 

With write cache enabled, writes are mirrored to the write cache memory in the other storage processor over the CLARiiON Messaging Interface (CMI). The CMI is a Fibre Channel based link and operates at either 100MB/sec on FC series or 200MB/sec on CX series systems. Each storage processor also includes a TCP/IP connection that is used for configuration and management of the storage system.Each storage system ships with a complete copy of FLARE software installed on the first four disks on back-end loop 0. Disks 0_0 and 0_2 store mirrored copies of the software for SP A, and disks 0_1 and 0_3 store mirrored copies of the software for SP B. When you power up the storage system, each SP boots and executes FLARE software.

Salient Features :-

# Storage processors are configured in pairs for maximum availability.

# One or two processors per Storage Processor board.

# Two or four Fibre Channel front-end ports for host connectivity

– 1Gb or 2Gb.

– Arbitrated loop or switched fabric.

# Dual-ported Fibre Channel Disk drives at the back-end.

– Two or Four Arbitrated Loop connections.

# Maximum of 4GB of memory per SP.

– Write Cache is mirrored between Storage Processors for availability using the CMI 

(CLARiiON Messaging Interface)

– Write Caching accelerates host writes.

# Ethernet connection for management.

An Introduction To Clariion Architecture.

The CLARiiON storage system is based on a modular architecture. The concept is “buy and build as you go”. The first building block of the architecture is the Disk Processor Enclosure, or DPE. The DPE houses the storage Processor(s) and the first Fibre Channel Arbitrated Loop (FC-AL) disks. Some older CX series arrays use the DPE2 which includes, in addition to the storage Processors, up to 15 2Gb dual-ported FC drives. Newer CX series, namely the CX300, CX500, and CX 700 can use a DAE2P, which provide for better fault isolation and reliability on the back end (BE) loops. The FC series DPE contains up to 10 1Gb dual-ported FC drives. In order to expand the capacity of the storage system, Disk Array Enclosures (DAEs) are interconnected using Link Control Cards (LCCs).The module architecture allows the customer to add drives as needed to meet capacity requirements.When more capacity is required, additional disk array enclosures (DAE or DAE2) containing disk modules can be easily added. LCC or Link Control Cards are used to connect shelves of disks. In addition, the LCC monitors the FRUs within the shelf and reports status information to the storage processor. The LCC contains bypass circuitry that allows continued operation of the loop in the event of port failure.

Newer CLARiiON arrays have two processors per Storage Processor, and do not use a DPE. Instead, it utilizes an SPE or Storage Processor Enclosure. The SPE does not contain any disk modules, so it must have at least one DAE2 and a maximum of 16 DAE2s. CLARiiON Architecture is based on intelligent Storage Processors that manage physical drives on the back end and service host requests on the front end, be it Fibre Channel or iSCSI protocols. Depending on the module, each Storage processor includes either one or two CPUs. Storage Processors communicate to each other over the CLARiiON Messaging Interface (CMI). Both the front-end connection to the host and the back-end connection to the physical storage is 2Gb Fibre channel.

What is a metaLUN in EMC Clariion ?

A metaLUN is created by combining 2 or more LUN's.MetaLUNs are supported only on CX-Series storage systems.The metaLUN feature lets you dynamically expand the capacity of a single LUN (base LUN) into a larger unit called a metaLUN. You do this by adding LUNs to the base LUN. You can also add LUNs to a metaLUN to further increase its capacity. Like a LUN, a metaLUN can belong to a Storage Group, and can participate in SnapView, MirrorView and SAN Copy sessions.During the expansion process, the host is able to process I/O to the LUN or metaLUN and access any existing data on the Base LUN. It does not, however, have access to any added capacity until the expansion is complete. Depending on the operating system, a reboot of the host or the use of a disk expansion utility, such as diskpar in Windows, may be used to claim the additional space. Each set of striped LUNs is called a component. All metaLUNs contain at least one component which includes the base LUN and one or more LUNs. Any data that gets written to a metaLUN component is striped across all the LUNs in the component. You can expand a LUN or metaLUN in two ways — stripe expansion or concatenate expansion. A stripe expansion takes the existing data on the LUN or metaLUN, and restripes (redistributes) it across the existing LUNs and the new LUNs you are adding.The stripe expansion may take a long time to complete and will affect performance while the expansion is in process . A concatenate expansion creates a new metaLUN component that includes the new LUNs and appends this component to the end of the existing LUN or metaLUN. There is no restriping of data between the original storage and the new LUNs. The concatenate operation completes immediately.

Salient Features :- 

# A metaLUN is created by combining LUNs

– Dynamically increase LUN capacity

– Can be done on-line while host I/O is in progress

– A LUN can be expanded to create a metaLUN and a metaLUN can be further expanded by adding additional LUNs

– Striped or concatenated

# Data is restriped when a striped metaLUN is created

# Appears to host as a single LUN

– Added to storage group like any other LUN

– Can be used with MirrorView, SnapView, or SAN Copy

# Supported only on CX family with Navisphere 6.5+