IBM Adds Speed, Flexibility To Big Data Storage
By Deni Connor
InformationWeek
IBM claims that it has shipped over 5,200 XIV Storage Systems and 384 petabytes of capacity since it acquired XIV in 2008. The product is targeted to big data, virtualized data centers, and cloud deployments.
An XIV Storage System consists of modules, each containing a multi-core CPU, RAM, and disk drives, that work in parallel to service I/O. Each module is linked with InfiniBand, and software running on the modules allows each to act as an independent storage system and handle its own caching and snapshots.
With the addition of an SSD cache, the company also claims that the XIV can perform at 160,000 I/O per second with 180 disk drives and 79 terabytes of capacity.
[ Sure, SSDs are expensive--but are they economical, too? See Solid State Storage Can Save You Money. ]
In IBM's implementation, the SSDs are used as a secondary cache between the hard disk drives and the DRAM cache. Data is always placed in both cache units and disk access is avoided when data resides in the SSD cache. The SSD cache needs to be at least an order of magnitude bigger than the DRAM cache to be effective.
In host reads, the XIV checks for data in SSD cache. It a hit is made, it re-routes the requests to the SSD for quick retrieval of the data. If not hit, it forwards the requests, unmodified, to the drives. Upon return, it copies the pages into an internal buffer. When the buffer fills up, it is de-staged to the SSD, sequentially, and sequential I/O detection bypasses the SSD to avoid cache poisoning.
In host writes, all writes go to the DRAM cache, and as they are de-staged from the cache, they are synchronously staged onto the SSD buffer. When the buffer is full, the data is written to the SSD.
The SSD cache, available on all Gen3 XIV systems, consists of a 400-GB SSD on each module and 6 TB of cache per rack. The SSD is housed in a PCI caddy in the rear of the module and can scale from 6 to 15 SSD drives.
The company also added the ability to mirror data between previous XIV storage systems and the new Gen3 systems. This allows customers to migrate data off old arrays and onto the new arrays and to then repurpose their older XIVs as disaster recovery backup systems.
IBM also added to the remote monitoring capabilities of the XIV by allowing iPhone users to manage the systems from anywhere. The app will be available for free download later this month. IBM previously released the XIV Mobile Dashboard for iPad users last year.
The company also promised IPv6 support in the first half of this year, followed by USGv6, the government implementation of IPv6, later in the year.
Deni Connor is founding analyst for Storage Strategies NOW, an industry analyst firm that focuses on storage, virtualization, and servers.
IT's spending as much as ever on disaster recovery, despite advances in virtualization and cloud techniques. It's time to break free. Download our Disaster Recovery Disaster supplement now. (Free registration required.)
Federal agencies must eliminate 800 data centers over the next five years. Find how they plan to do it in the new all-digital issue of InformationWeek Government. Download it now (registration required).
 |
To upload an avatar photo, first complete your Disqus profile. | View the list of supported HTML tags you can use to style comments. | Please read our commenting policy. |
Subscribe to RSSVirtual Infrastructure Reports
Informed CIO: VDI Snake Oil Check
You won't lose your shirt on a desktop virtualization initiative, but don't expect it to be simple to build or free of complications. This report examines the three biggest problems when developing a business case for VDI: storage costs, ongoing licensing, and the wisdom of prolonging the investment in PC infrastructure.
Fundamentals: Next-Generation VM Security
Server virtualization creates new security threats while turning the hypervisor into a network black hole, hiding traffic from traditional hardware defenses -- problems a new breed of virtualization-aware security software tackles head-on.
Delegation Delivers Virtualization Savings
IT can't-and shouldn't-maintain absolute control over highly virtualized infrastructures. Instituting a smart role-based control strategy to decentralize management can empower business units to prioritize their own data assets while freeing IT to focus on the next big project.
The Zen of Virtual Maintenance
Server virtualization has many advantages, but it can also lead to chaos. Many organizations have unused or test VMs running on production systems that consume memory, disk and power. This means critical resources may not be available in an emergency: say, when VMs on a failed machine try to move to another server. This can contribute to unplanned downtime and raise maintenance costs. Easy deployment also means business units may come knocking with more demands for applications and services. This report offers five steps to help IT get a handle on their virtual infrastructure.
Pervasive Virtualization: Time to Expand the Paradigm
Extending core virtualization concepts to storage, networking, I/O and application delivery is changing the face of the modern data center. In this Fundamentals report, we'll discuss all these areas in the context of four main precepts of virtualization.
Virtually Protected: Key Steps To Safeguarding Your VM Disk Files
We provide best practices for backing up VM disk files and building a resilient infrastructure that can tolerate hardware and software failures. After all, what's the point of constructing a virtualized infrastructure without a plan to keep systems up and running in case of a glitch--or outright disaster.