Molina Healthcare is an experienced implementer of the latest virtualized infrastructure, and its quest for the best continues with an experiment in the area known as converged infrastructure. The NYSE-traded healthcare company is installing “hyper-converged” infrastructure from Nutanix to work alongside its existing VCE converged infrastructure known as Vblocks.
Vblocks are considered in some quarters to be one of the most painless, if pricey, ways to install servers, storage, and networking, already loaded with VMware virtualization software and ready for compute-intensive tasks. Vblocks come in preconfigured racks, and they're produced by VCE, a consortium of Cisco and EMC, with VMware also as an investor.
Nutanix is one of a small group of suppliers of the recently dubbed "hyper-converged" infrastructure, where the base construction unit is a 2u appliance going into the rack, not a full rack itself. Nutanix provides embedded software that makes the unit aware of other Nutanix appliances around it and links their hard drives together into a combined storage network, also managed by Nutanix software.
Nutanix combines Intel Ivy Bridge blade servers with plentiful solid-state disk space attached to the system bus, allowing much of the data used in a given business day's processing to be found without frequent I/Os to hard drives. The NX-7110 appliance, for example, ships with up to 256 GB of RAM, plus two 400 GB solid state disks, supporting a total of two Intel Ivy Bridge 2680 sockets, with 10 cores each.
[Want to learn more about VMware's hyper-converge infrastructure? See VMware Debuts EVO: Rail.]
Molina is a Fortune 500 company and processor of claims from 5 million low-income and underinsured healthcare users in 15 states. In effect, it's processing claims for families that receive their care through Medicaid, Medicare, and other government-funded programs. As such, it relies on complex claims-processing systems that need to be aware of the rules and regulations governing programs in those 15 states.
To do so, it runs much of its internal operations as a private cloud, hosting 3,500 to 4,000 virtual machines on its converged infrastructure at any one time. Ryan Blace, associate VP of infrastructure-as-a-service, said Molina installed 10 Vblock clusters four years ago and continues to make use of them. But six months ago, it installed its first Nutanix appliance cluster as a replacement for one of the Vblock clusters.
"Nutanix is much simpler to operate," Blace said. "It has all the enterprise features of a storage array," without IT doing anything to configure an external storage array. Setting up a storage array is generally labor intensive, since a network manager needs to configure and connect it to the SAN. Nutanix units are able to form an equivalent of an array, using disks already on the rack where the Nutanix 1u and 2u-sized appliances are running.
"Each compute node can see all the storage that's available" over a standard NFS network file system, said Blace, even though much of it is built into other appliances. Nutanix produces an appliance with less compute power on board but more disk. Its NX-1050 appliance, for example, comes with 12 cores, but four 1 TB hard drives. The more compute intensive NX-7110 has a single 1 TB drive. The idea is that by mixing and matching appliances in the rack a Nutanix user can balance compute and storage needs without installing and configuring external storage arrays or a SAN.
Furthermore, a hard drive on a node in the cluster can fail, and its workload gets shifted over to other drives. When IT replaces the failed unit, Blace said it's recognized as a new drive on the network and reintegrated into the infrastructure. The system's load balancing is good enough for Molina to set a threshold of 75% utilization of memory and 75% utilization of storage as an operational parameter, and hit it frequently. Blace said it's hard to set the usage of both resources that high in data centers where the IT staff is doing the integration of the virtualization components.
In terms of cost, Blace is finding Nutanix and VCE clusters, each running 350 to 400 virtual machines, are comparable. The savings will come in lower operating expense. For example, Blace likes that he can upgrade a Nutanix cluster by just adding appliances to the rack, without costly testing that eats up staff time. While he wants more thorough reporting on Nutanix versus Vblocks before comparing operational expense, he thinks shifting to Nutanix will help Molina keep its IT staff of 250 flat, instead of growing at the 25% rate at which the business is expanding.
Blace sees some costs savings coming from the Nutanix systems ability to manage storage and the storage network by itself, including data deduplication and replication. With its Vblocks, Molina is paying for add-on software to manage storage data, Blace said, including such items as EMC's Storage Resource Manager, EMC Recovery Point, and EMC Power Path dual-data pathway software.
Blace said the SSD-intensive Nutanix design also makes for low-latency I/Os when data is needed by the CPUs. There are no calls to disk that go out from the server's host bus adapter through a storage array's controller to the hard drive and back again. The Nutanix system manages the data in the SSDs, using them as a large cache, but only starting to move less frequently used data to disk once the SSDs are nearing their limit. That keeps as many I/Os as possible free of calls to a hard drive, even though it's nearby and onboard the rack.
Molina has looked at a number of hyper-converged infrastructure providers. There are a handful of suppliers, including VMware and its EVO: Rail.
Blace wouldn’t put a time frame on a transition to hyper-converged infrastructure and added that Molina will continue to use Vblocks. But hyper-converged infrastructure from Nutanix “is very promising.”
(Editor's Note: This article was updated on Jan. 28 to reflect changes communicated in a followup interview with Molina Healthcare's Ryan Blace.)
Attend Interop Las Vegas, the leading independent technology conference and expo series designed to inspire, inform, and connect the world's IT community. In 2015, look for all new programs, networking opportunities, and classes that will help you set your organization’s IT action plan. It happens April 27 to May 1. Register with Discount Code MPOIWK for $200 off Total Access Conference Passes.Charles Babcock is an editor-at-large for InformationWeek and author of Management Strategies for the Cloud Revolution, a McGraw-Hill book. He is the former editor-in-chief of Digital News, former software editor of Computerworld and former technology editor of Interactive ... View Full Bio