Web 2.0 infrastructures ready to move to the corporate data center
Even if you don’t spend a lot of time on Facebook or Twitter, you probably know they built huge, valuable businesses without spending a lot on proprietary hardware and software. They reliably serve hundreds of millions of users with only low-cost processors, storage and networks. Their secret sauce: Virtualization and management software that enables this commodity gear to deliver scalability, reliability and flexibility that would otherwise cost millions from a mainstream vendor.
Why, you’ve probably asked, can’t you get the same results in your own data center? On the server side, you’re increasingly able to, thanks for the likes of VMware, Microsoft Hyper-V and Citrix XenServer. On the storage side, you’re much more limited. The more enterprise-level features you need (such as storage tiering, fault tolerance or massive scalability) the more you are forced into expensive vendor-specific hardware.
All that is changing, thanks to advances in data management software. These new capabilities will allow Flash, SSD and traditional hard drives, combined with commodity servers, to deliver the massive, cross-workflow scalability once available only to the start-ups of the world. Here’s what’s coming.
Flash/SSD: Not Just for Storage Arrays
As prices for flash (the memory within solid state drives, or SSDs) have fallen and capacities have risen, just about every storage vendor has begun offering SSDs along with hard disk drives as a “tier” of storage available in their arrays. Because SSDs are still much more expensive and store less data than traditional disk drives, they are usually used to cache only the organization’s most performance-sensitive data.
However, this caching saps utilization and performance due to the overhead of moving data from legacy storage to the SSD devices in an array or Flash in the server. It also wastes large amount of storage space, because cached data must be stored twice (once on primary storage, and again in cache.) Caching wastes yet more space because any changes to data must also be stored in both locations. The constant cycle of writes and deletes shortens the life of flash memory, while its placement on a single server makes it a single point of failure unacceptable for enterprise-grade applications. It also creates silos of application data that make it more difficult for users, customers and business partners to get the information they need to grow the business.
This is not delivering the value promised by all the hype around flash. But there’s another approach that will drive quantum reductions in storage cost, improvements in performance and in the scalability and reliability organizations require. That is to deploy NAND memory not within traditional storage arrays, but in the server stack. This Server-side scale-out storage approach not only boosts performance, but (combined with data management software such as our Melio clustered file system) delivers enhanced availability and scale-out capability to enterprise and applications at dramatically lower cost
Of course, the “intelligent management layer” must support flash and SSD along with conventional hard drives to meet the cost to performance to capacity ratios expected by customers. That’s exactly what our recently announced Latency Targeted Allocator module does, adding support for server-side PCI Express flash cards and solid state drives to the traditional hard drives already supported by our Melio data management platform. This will increase the performance of your storage platforms by 4 to 10 times, while dramatically reducing capital and operational costs without sacrificing capacity, reliability, scalability or SLA’s.
We’ll explain more details of LTA in later posts, along with case studies of soon-to-come enhancements to Melio that will bring exponentially greater scale out, reliability and low cost to multiple enterprise workloads.
For now, suffice it to say that the basic “recipe” of low-cost hardware topped with a virtualization and management layer isn’t just for startups anymore. Mainstream enterprises have the same needs for reliability, scalability, performance and dramatically lower costs. So far, Google, Amazon and other Web services providers have made such capabilities available only to customers of their own public cloud offerings.
We’re working hard to make these capabilities available to you, the enterprise customer.
Let us know what “Web 2.0” use cases you need addressed, drop a line or if you have a short-need we can help you with.
Follow me on twitter @mvmsan
Experience for yourself why OVER 700 customers globally have chosen Sanbolic over more costly and complex alternatives. Melio for FREE