Cloud Security FUD Drives Genomics Industry towards Cloud-in-a-Box, Part 2

Accessibility and cloud security concerns push development of on-premises cloud alternatives for the life sciences and genomics industries.

By Joe Stanganelli | Posted May 21, 2014
Page of   |  Back to Page 1
Print ArticleEmail Article
  • Share on Facebook
  • Share on Twitter
  • Share on LinkedIn

Click here for Part 1.

"It's an Xbox for DNA," chuckled Ralph Van Der Pols as he introduced his company's product to me on the trade show floor at the Bio-IT World Conference and Expo in Boston.

The product, known as the "Vault," is a Next-Generation Sequencing (NGS) appliance made by Netherlands-based biomedical analysis firm Genalice. Genalice unveiled the Vault on April 29, the first day of the Bio-IT World Conference. The Vault, which comes with specialized GUI, storage, database, and monitoring software, is little more than a turnkey console running on a basic dual Intel Xeon E5-2600 Series CPU.

But impressive as it appears, the Vault is not an entirely unique product. The health and life sciences sector, ever hungry for bigger and more accessible data and ever more concerned about cloud security and regulatory compliance, is finding alternatives to the cloud in on-premises, cloud-independent turnkey consoles: "cloud-in-a-box" solutions.

On-premises cloud alternatives solve cloud security and functionality problems

Seven Bridges Genomics debuts cloud alternative for the life sciencesKate Blair, director of product management for Seven Bridges Genomics, likens her company's new turnkey NGS console to an Xbox, too. Custom-developed by Massachusetts consultancy BioTeam (in collaboration with Intel) on the BioTeam SlipStream Appliance, the Seven Bridges Appliance debuted at Bio-IT World at a booth directly across from Genalice's. The Seven Bridges Appliance is slated to ship this summer.

Stan Gloss, CEO of BioTeam, thinks that the Xbox/gaming console analogy is apt. Because of console standardization in the gaming industry, "[the] cost of games come[s] down and the quality goes up," Gloss told me. "There's a reason why there's a demand [for] standardized appliances. Standardization helps markets mature."

Blair contends it's simpler than that. "[S]ome people just like buying boxes," she said with a smirk.

It's funny because it's true. Blair explained to me that, even aside from those who naturally shy away from the cloud due to cloud security risks or other reasons, there is still a very real – and very justified – demand for these turnkey cloud alternatives when it comes to cost-effectiveness, depending upon a client's on-premises infrastructure and needs. In particular, Blair emphasizes that a scalable turnkey console may complement an organization's pre-existing local storage better than a cloud solution would. (Blair playfully described this scenario as having "boxes for your boxes.")

Other Bio-IT World attendees seem to agree with Blair's assessment.

Dave Henry, an Accunet Solutions storage and virtualization executive with more than 25 years of IT experience, told me that the major IT problem he most commonly sees is clients "not taking best advantage of the storage they've already connected."

On-premises appliances beat out cloud infrastructure for the genomics enterprise

"[I]nfrastructure is…a money-solvable problem," Sebastian Wernicke, director of Seven Bridges, told Bio-IT World attendees in a sponsored presentation. Storage, processing, and computation capabilities, Wernicke posited, are all necessary considerations when it comes to big data, particularly in how they work together.

"It is, fortunately, pretty predictable how much you will need," says Wernicke.

Indeed, Blair pointed out that if the amount of analysis a client conducts is fairly limited – and, therefore, doesn't demand the full computing power of the cloud – it may make more sense (and save more money) for that client to go with a standardized turnkey cloud alternative on-premises. (Blair and her colleagues estimate that the Seven Bridges Appliance can analyze one genome a day.)

That may be a big "may," however.

R. Mark Adams, CIO of Good Start Genetics in Cambridge, Mass., told attendees in a pre-conference workshop on cloud infrastructure versus on-premises computing for NGS, "In general, the only way it's usually cheaper to buy the equivalent hardware all-in is if you're gonna run it [at a] 90-percent load."

Cloud security and compliance concerns in genomics

Nevertheless, there are other costs to consider besides the obvious, upfront ones. Benjamin Breton, a bioinformatics software engineer at Good Start Genetics, pointed out at the same workshop that, when it comes to health and life science regulations like HIPAA, "Everything needs a checklist. If you deviate from that checklist, you need to document it[.]"

Breton argued that, accordingly, if you move your data from on-premises storage to the cloud, then there is a good chance you will need to re-document everything. What's more, Breton says, "[e]ncryption is a big deal" under HIPAA; to remain in compliance, a health or life sciences firm will probably need to pre-encrypt its data before moving it to the cloud to address cloud security risks to protected information.

And then there is the need for lawyers, business associate agreements (which can include "gotchas," Breton warned) with cloud partners, security audits, ensuring proper API support, and all the other usual cloud nuisances. (This, of course, is not to pooh-pooh the cloud in general, but merely to point out that these things add to the cost of a cloud investment.)

Supplementing on-premises appliances with cloud computing

DRAGEN chip helps provide cloud alternative for genomicsIn any event, Adams's averment may depend upon the particular on-premises solution. The Genalice Vault, for instance, converts data into a proprietary (yet convertible) file type that can save up to 100 times the storage space compared to standard genome sequence data file formats. KBioBox, a Worcester, Mass., startup, offers similar cloud-free consoles that reduce storage footprints while extracting actionable genomic information a minimum of 60 times faster than standard enterprise cloud configurations.

Yet another company at Bio-IT, Edico Genome, announced its new DRAGEN chip. The DRAGEN chip boasts the ability to independently offer local genomic mapping and alignment at more than 150 times the speed of 50+ high-end servers that have 12-core 2.7GHz Intel processing and 96 GB of DRAM. Edico anticipates a June release for DRAGEN, which is currently in the prototype stage.

"We're not trying to say you don't need the cloud," Pieter van Rooyen, Edico Genome's CEO, told me during an interview. In fact, his company offers cloud analytics to complement DRAGEN's on-premises abilities. Rather, Edico's local solution, van Rooyen explained, can simply do the heavy lifting locally.

"The answer can go to the cloud," said van Rooyen. "You don't need all these extra[s]."

Still, with turnkey cloud alternatives such as these, customers typically explicitly pay for extras to go with their high-end hardware. Blair was quick to note that whether a Seven Bridges customer chooses to go to the five-year-old company's AWS cloud platform or stay local, that customer will still have full access to the company's bioinformatics support team. "We can even help you select the tool…for your experiment," she says.

Of course, the choice between cloud computing and on-premises processing is not mutually exclusive. Cloud security is a worry, but so are the scalability and cost of on-premises devices. Local processing consoles that can work independently of the cloud or be cloud-enabled offer the best (and worst?) of both worlds.

Regardless, any decision about whether to go to the cloud or the "anti-cloud" (or both) must involve serious cost-benefit analysis.

Henry, grizzled IT veteran that he is, summed up the question more cynically: "Which headache do you want?"

Comment and Contribute
(Maximum characters: 1200). You have
characters left.
Get the Latest Scoop with Enterprise Networking Planet Newsletter