Achieve High-Performance Data Storage Migrations At Top Speed With Cirrus DMS
By Wayne Lam, CEO, Cirrus Data Solutions
Special guest blogger Wayne Lam shares expert insight about Cirrus Data Migration Server, the first zero-downtime disk-block level solution for legacy FC SAN data migrations supporting all applications, physical or virtual hosts, storage technologies, and destinations—local, remote, or cloud. Discover how Cirrus helps customers plan better strategies, achieve top performance, improve quality, and get up to 8X faster data migrations.
It is my honor to be receiving so much attention lately, including blogs, videos, and other demonstrations of our products and services. In this special partner blog with CDI, I would like to immediately establish in your mind two main ideas.
First, our technology and experience in migrations can help every customer situation. I’m often presented with a customer scenario that may seem difficult or impossible from the customer’s point of view; however, without sounding over-confident, we want to comfort all customers with the assurance that we have a solution for you. To all R&D, healthcare, finance, media, and even the largest multi-national firms with unique storage policies/requirements, and different technologies—we welcome your toughest data migration goals! And working together, Cirrus and CDI can support any host, application, or storage configuration, and we mean any… VMware, Windows, Linux, AIX, Oracle, SQL Server, IBM, SAP, Dell EMC, HP, Exchange, Pure, NetApp, HDS, Fujitsu, and more.
Second, we take planning, performance, and quality of service (QoS) very seriously. At Cirrus Data, our people thrive inside a culture of innovation and exude a spirit of sharing ideas to solve challenging problems. Companies rely on us to architect their new storage infrastructures and platforms. During migration, we set high expectations for QoS metrics that include application performance, speed, and utilization. We support our partners to advise you along every step of the way in even the most demanding online data migrations running 24×7 so that you receive the best QoS and maximum storage bandwidth utilization without sacrificing performance.
Data Migration Challenges
To illustrate these themes, I’ll share a recent success story with you where our customer enjoyed a worry-free solution that conformed to a unique, no-touch storage policy. After a European bank merger, our client, a major financial group, wanted to consolidate two large data centers in Vienna and relocate to Italy. Because they had a unique “no-touch” migration policy, they had concerns about security and efficiency. Four old EMC arrays with 64 active FC ports were in active production and needed to migrate to an IBM SAN Volume Controller (SVC) storage farm. They had a mix of operating systems, including different versions of AIX, and more than 20,000 LUNs on those four arrays. In total, it was over 2,000 initiators on 500+ hosts. Due to the mission critical nature and sensitivity of each application being migrated, it was necessary to have a migration strategy that could “divide and conquer.” This strategy allowed for 100% isolation of the apps using separate migration groups to ensure no performance hits or issues with other applications.
Typically, our Data Migration Server (DMS) requires no zoning or configuration changes. We integrate the DMS into existing fabrics using Logical Insertion that is pretty much plug-and-migrate. However, in order to divide and isolate each group of apps, we collaborated with a migration partner on the following strategy:
- Designated controller ports on the source storage arrays exclusively for migration.
- Use the “Dedicated Migration Path” insertion method to seamlessly insert DMS appliances into the live SAN Fabric without any downtime.
- Deploy multiple pairs of DMS-8000s to split the operation into multiple workloads, where each batch targets a group of hosts using a particular application to manage the very large number of client hosts and associated LUNs being migrated.
- Set the iQoS settings appropriately for each workload such that migration throughput can be maximized with precisely negotiated impact level to the host I/O, based on day of the week and time of day, on an hour-by-hour basis.
We migrated with zero downtime and delivered on the one-time insertion requirement without making any changes to existing production FC ports. Our TCP/IP remote replication feature with compression and encryption resulted in an average of 5 times bandwidth gain between the two countries.
The performance was spectacular—with 5:1 compression, we achieved throughput rates over 500 MB/sec for each gigabit link with zero complaints from host application owners on application impact. When production storage is migrated while the workload is still active (known as “online storage migration”), traffic jams on the storage network often occur and inevitably will impact application performance. This is especially true for other migration tools that are host-based, especially in large-scale migrations of 100 terabytes or more. Other tools that are repurposed for migration such as third-party disaster recovery tools, built-in logical volume manager (LVM) mirroring, and VMware Storage vMotion often show a significant, negative impact to application host performance.
We can attribute the excellent performance results to the Intelligent QoS (iQoS) feature of DMS. SAN traffic typically has numerous small gaps that could be used for migration at full speed without impacting the host. By making use of these gaps for reading the data, DMS efficiently uses all available bandwidth and maximizes migration rates with little or no impact on application traffic. The DMS iQoS feature monitors the data path constantly, looking for gaps. We can actually see the real-time traffic conditions. When congested, we can pause the migration, and when traffic subsides, we automatically resume migration at full speed, without negatively impacting the host application traffic.
You get the fastest data migrations at the highest possible production storage QoS on your SAN.
Want to Learn More?
To learn more about the customer success story and performance details referenced in this article, visit the following links on the Cirrus website:
About Cirrus Data
Cirrus Data Solutions, Inc. is a privately-held firm providing enterprise-class technology solutions for automated data management and optimization. Founded in 2011 by brothers Wai and Wayne Lam, storage experts with decades of innovation and success in backup and data recovery, the company employs an elite team of R&D, engineering, and management professionals to drive its vision. With 26 patents, seven of them centered on its groundbreaking Transparent Datapath Intercept (TDI) technology, Cirrus Data develops Data Migration Server (DMS) and Data Caching Server (DCS). Both of these market-leading plug-and-play solutions install in minutes for fast, simple, uninterrupted data management with zero downtime to help customers achieve their data optimization goals.