FAQ: SAP's HANA in-memory computing platform

Here's a look at some questions surrounding a product that will "revolutionize" SAP's portfolio

This week's SAP Sapphire Now conference marked about a year since the company launched its HANA (High Performance Analytic Appliance) in-memory computing platform. Since then, SAP has done its best to keep HANA in the news, bringing products and partnerships to market quickly and announcing many future plans.

Over time, HANA will "revolutionize" the company's entire software portfolio, SAP CTO and executive board member Vishal Sikka said this week. With general availability of HANA coming next month, that journey is just beginning.

[ Discover what's new in business applications with InfoWorld's Technology: Applications newsletter. | Get the latest insight on the tech news that matters from InfoWorld's Tech Watch blog. ]

Here are answers to some of the main questions around HANA, based on interviews and Sapphire speeches by Sikka and SAP co-founder Hasso Plattner.

What is it?

HANA, which will be available in appliance form from a number of hardware partners, places data to be processed in RAM instead of reading it off disks, adding a performance boost.

HANA is built on a superset of technologies with a long history at SAP, including the MaxDB database and TREX in-memory engine, Sikka said in an interview.

Data held in memory by HANA is backed by a persistence layer that logs transactions and incorporates savepoints that create images of the database, according to an SAP document. This allows it to be restored in the event of a power outage or other disruption.

HANA is compatible with any BI (business intelligence) application that supports common query languages like SQL and MDX, according to Sikka.

What can it be used for?

SAP has initially focused on HANA's ability to support real-time analytics, especially as part of a series of specialized applications that target discrete business problems. One of the first such products is Strategic Workforce Planning, which companies can use to figure out the economic and logistical ramifications of making big shifts in staffing.

HANA can also handle the transactional workloads of ERP (enterprise resource planning) applications like its Business One, Business ByDesign, and Business Suite products. All of those products are expected to gain HANA support, but the enterprise-grade Business Suite is expected to take the longest time.

In the meantime, Business Suite customers looking for an alternative to Oracle or other databases should look to Sybase ASE, which SAP is just about done porting over.

How scalable is it?

HANA is lightweight enough to run "a company" on a Mac Mini server, but it is also capable of scaling out massively, Plattner said during his Sapphire keynote.

But the company has also created a HANA system with more than 1,000 cores, 16TB of RAM, and 64TB of solid-state storage. HANA scales linearly, meaning that if you need more cores or memory, you simply add more nodes, according to SAP.

HANA also employs advanced compression techniques, allowing much more data to be crammed into memory.

How much does it cost?

This one is complicated and could be subject to significant change. SAP has said HANA appliances will be available in "small, medium, and large" sizes. There's a standard bill of materials SAP has developed for HANA machines, according to Sikka. That should provide some parity across various hardware partners' products.

SAP's current approach to licensing HANA is based on the amount of data processed, but the company is looking at several other ways, Sikka said in the interview.

SAP is also planning to price HANA based on the amount of actual business value or cost savings it can provide customers.

"It's not the number of processors, that model. We want [customers] to use as many processors as possible," he said.

While this may seem to be a more nebulous way of arriving at a deal, SAP has practice at it through its Value Engineering program, Sikka and Plattner said.

"You basically have to predict what will happen [in terms of savings], and then you have to fulfill it," much like a service level agreement, Plattner said.

Overall, more information about pricing could emerge in June, when HANA is set for general availability.

What is the HANA AppCloud?

During his Sapphire keynote, Sikka gave a brief preview of the upcoming HANA AppCloud. This is "an application environment for native on-demand applications based on SAP's in-memory technology," and will allow customers to take advantage of in-memory technology along with the usual cloud-related benefits of scalability and flexible deployments, according to a spokeswoman.

SAP's Business Intelligence OnDemand, Carbon Impact and Sales & Operations Planning applications are among those that will be available.

The HANA AppCloud is now in "pre-beta" and goes into "ramp-up" release mode in October.

Other aspects of the platform still aren't public, such as whether end-user companies as well as ISVs (independent software vendors) will get to use it, and if so, whether they will be allowed to run applications in production form. Even if the answer to that is no, the AppCloud could give enterprise IT shops a chance to test-drive the technology before making a major investment in HANA.

What hardware platforms does it support?

A wide range of vendors have signed on to sell HANA appliances, including IBM, Fujitsu, Dell, Cisco, and Hewlett-Packard. Oracle is a notable exception, but not a surprising one, given the competition HANA appliances stand to give to its Exadata machine.

But in one sense, SAP has kept HANA hardware support constrained, by working closely with Intel to target only its x86 chips. This has given SAP HANA teams more agility and sped up development, since only one platform needs to be supported, Sikka and Plattner said.

The partnership is also helping SAP squeeze more power out of HANA thanks to Intel-specific optimization.

SAP had been running HANA on eight-core Nehalem chips from Intel, Sikka said. It recently tested Intel's new 10-core Westmere-EX processor with HANA and found that queries were 37 percent faster, he said. "Eight cores is only 25 percent less than 10 cores, so how could this be 37 percent? It's because of the optimizations that are Intel-specific."

While the x86-only plan could change down the road, it's in place for the foreseeable future.

"Our highest priority now is speed to market, to help customers develop much faster," Plattner said. "All of them say to me, 'We need this, we want this.' That's not saying other chips are not good. They could even be better. It's all about speed."

How often will HANA be upgraded?

Years can go by between a major release of a database. HANA will take a much different approach, one that is easier on customers, Plattner and Sikka said in the interview. There will be no version numbers, only service packs.

"We stopped this 'release thing.' We will ship additions and they have to be fully compatible," Plattner said. "Whether it's on-site or in the cloud, continuous improvement without disruption."

The updates may be small or large, but the focus on non-disruption will persist, he said. "Salesforce.com can do [this style of development], Facebook can do it, Google can do it, we can do it."

Chris Kanaracus covers enterprise software and general technology breaking news for The IDG News Service. Chris's email address is Chris_Kanaracus@idg.com

Copyright © 2011 IDG Communications, Inc.