By: Eric Maille, Enterprise System Engineer, EMC

The virtual Datacenter.fr - En route vers les infrastructures de nouvelles générations


A large service provider recently signed a multi million-dollar deal with EMC SDS Solutions.  With this type of solution, the customer reduced its storage costs by 25%. This reference confirms the trend that customers are moving more and more towards Software Defined solutions, which aligns with the predictions from IDC, Gartner, Wikibon ....Here are some of my recaps and opinions on ScaleIO, based on what I’ve seen in the industry.


ScaleIO is one of the core Software Defined technologies at EMC. This innovative solution and the disruptive approach require a change in our traditional mindset. ScaleIO is not meant to replace all the workloads in the IT environments, but it’s key to accommodate Next-Gen applications.
 
ScaleIO’s principle is quite simple: a set of x86 servers is aggregated with local drives, which creates a virtual SAN. All the features of traditional storage arrays are preserved and offered at at a lower cost with a simplified management.

ScaleIO offers very high performance and high scalability while reducing management complexity and costs.

 

What have I seen as the top value proposition of ScaleIO?
  



The current challenge of traditional architectures (so-called second generation) based on SAN + Servers + Storage often represent a heavy burden for IT managers. In many companies, managers consider that their IT is extremely expensive and offers a poor level of service. They experience high TCO, complex management, low reactivity for resources provisioning and limited scalability, and eventually blame on the IT for not being able to serve their needs such as Big Data, Mobile, Social, the Internet of Things, and more. To address these challenges, it is necessary to rethink IT and make IT AGILE.


The underlying architecture allows for a simplification strategy and agility thanks to provisioning and de-provisioning on the fly. It’s no longer necessary to manage legacy equipment, Fabric SAN, HBA, nor to update the Firmware to make it compliant.... With ScaleIO, the TCO is reduced by 65% compared to a traditional SAN.


 

Its Scale OUT architecture offers very high scalability (thousands of nodes: up to dozens of PB and several millions of IOPS). It grows together with the application requirements and applications are closer to their storage. The more are added, the more the nodes increase performance and / or volume.
ScaleIO runs on servers (x86) which allows to standardize the IT environment and eliminates the need to manage proprietary solutions that would require specific training. 


What are the use cases I’ve seen for ScaleIO?


Dev Ops: ScaleIO solution allows you to add or remove capacity on demand with automatic data rebalancing, while preserving operations. This feature is ideal for environments that require Devops do proceed with frequent changes when switching between Test to Dev to Qual to Pre-prod to Prod.


Applications requiring high levels of performance: The scale-out side of the solution enables evolution as the application requirements change. The tests performed benefit from the SSD technology and can achieve unmatched performance levels. Example with 53 knots profile 4k: 



Eric1.png


Traditional applications that require high performance such as SAP, Oracle, real-time applications ... are a good use case.
 



Test/Dev: The Test / Dev is a dynamic environment that requires regular changes, such as add or delete operations. The elasticity of the solution is a good answer to these issues.
New apps or innovative projects: to start new projects without breaking the bank, or when a customer has no idea what the required performance or scalability should be. You can start small –with at least 3 nodes- and grow as the application evolves.


Technically, here is what you should know about ScaleIO:

ScaleIO Architecture

 

ScaleIO is built on 3 main components: MDM, SDS, and SDC.



• MDM (Meta Data Manager) configures and monitors the ScaleIO system. You can have 2 for redundancy (1 + Tie Breaker). Note that MDM is not in the Data Path. It helps monitoring and configuring the cluster, and contains information Mapping.


• SDS (ScaleIO Data Server) provides the abstraction layer , contributes to the storage pool and provides IO operations. SDS is installed on each participating server in the storage system.


•SDC: a driver that exposes the ScaleIO volumes to the applications that are hosted on the same server. This driver runs on the server and provides access to volumes through TCP/IP. In other words, the app communicates with SDC volumes through this driver.


Protection Domain, Storage Pool Set Fault.
When configuring a ScaleIO system, several elements are set up.


Eric2.png

• Protection Domain: a group of servers (or a set of SDS). An administrator can segment all SDSs in several protection areas. Rebalancing and redundancy features are established in the Protection Domain. Each application of each server can access all volumes based on its assigned PD.


• Storage Pool: a subset of the physical storage in a PD. Each drive is linked to a single storage pool. A volume is distributed through all the devices that are hosted on the same storage pool. It’s recommended to have one single type of storage within the same pool.


  • Fault set:

When writing, each block (or chunk) is mirrored to another node in the cluster to ensure redundancy in the event of a disk (or node) failure. A Fault set helps organize the distribution of copies of the nodes. A Fault set is a logical entity that holds a set of SDS within a protection domain. Fault set A can be defined as a set of servers that can fail simultaneously with no impact on production. ScaleIO never copes twice the same copy of data within the same Fault set.



Eric3.png

Advanced features
• Replication is performed by RecoverPoint
• OpenStack Cinder supported
• In VMware environment, ScaleIO is deployed as a VM SVM (ScaleIO VM)
• ScaleIO comes in 3 configurations:

o   Software installed on x86 servers

o   ScaleIO Node: Hardware + Software package\

o   VxRack: complete out-of-the shelf IaaS solutionEric4.jpeg.jpg

Eric Maillé |

http://www.thevirtualdatacenter.fr/WordPress3/

Enterprise Systems Engineer

EMC² France  |  River Ouest  |

80 quai voltaire |  95870 Bezons

Mobile : +33 615127016