In-memory computing: the SMB perspective
11th Jun 2013 | 16:30
Advanced software and cloud services are removing barriers to entry
Having gained traction in the larger enterprise in recent years, in-memory computing (IMC) is now becoming a viable option for small to midsized businesses (SMBs).
IMC involves storing critical business information in the random access memory (RAM) of dedicated servers, rather than in relational databases on traditional hard disks. It removes bottlenecks and positions data closer to the processor, allowing analysis to take place in 'real time'.
According to a recent Gartner report, the declining cost of RAM and maturing software systems are making it easier to SMBs to pick up on IMC.
"The industry has been through a period where large enterprises have used in-memory computing on a very experimental level," says Bo Lykkegaard, analyst at IDC. "The technology is now beginning to spread to smaller companies via different routes off the back of that."
A number of vendors targeting smaller businesses are beginning to integrate IMC technology into their analytics solutions, including Tibco Spotfire (in Spotfire5), QlikTech (Qlikview), and PowerPivot (Microsoft), which allows users to load data in-memory that can be manipulated using Excel.
"These vendors provide business intelligence tools that allow you to load in-memory data from your data warehouse or from various data sources," explains Gartner analyst Massimo Pezzini.
"While the data is in-memory you can analyse it using graphical or navigational tools that return results very quickly."
Pezzini adds that SMBs should not think in terms of how much data they have when considering an IMC solution, as a more important question lies in the frequency of reporting.
"You probably don't need in-memory computing if you only run reports once a week or month," he says. "However, you would need a data visualisation tool if you needed to analyse data in real time to identify opportunities and track profitability of your product line or customer activity."
Tableau Software, a provider of data visualisation solutions, offers an "in-memory Data Engine" which can extract data to bring it in-memory before executing fast query responses on hundreds of millions of rows, all performed on scalable commodity hardware.
"When users are able to get answers to their questions at the speed of thought they are able to ask more questions of their data, and safely explore different assumptions and perspectives," says Francois Ajenstat, Director of Product Management at Tableau.
"If people can make decisions based on facts and figures rather than gut feel then the quality of the decisions will increase which will result in greater business outcomes."
Innovating at speed
While SAP is not credited with inventing in-memory computing (it has been around since the 70s), it was the first vendor to ramp up marketing efforts around its own IMC technology when it launched HANA in late 2011.
SAP recently integrated HANA's technology, which allows businesses to run both analytics and transactions on a single database, with its BusinessOne management suite for SMBs. This removes the need for a separate data warehouse to run reporting tasks, thus lowering the potential total cost of ownership (TCO).
"Simplistically, HANA is an added-on appliance to BusinessOne, which allows the customer to pull data from the transactaional database and run analytics on top of it," says Rich Philips, UK Channel and Sales Director at SAP. "It's pretty low risk as you don't have to switch off the BusinessOne system while you're deploying and delivering the analytics alongside."
Philips adds that this could give SMBs a competitive advantage, providing the example of a retailer with six or seven stores that needs to find out whether a specific TV is in stock.
"If a customer enters a store and wants a particular TV that's not there, but is in another store, the SMB only knows the exact stock level of that particular item at that time," he says. "The current method to get an update would be to do an end-of-day stock check and submission to the back office, which could risk the customer going to that store to find out the TV has been sold."
"The benefit of IMC is that you don't have to wait for the transaction to complete until you can perform all of the analysis on it and get an update. That can help keep customer dissatisfaction at bay, and of course, loyalty in today's economy is very short lived."
For businesses unable to afford the on-premise route to deploying IMC, IDC's Lykkegaard says that cloud solutions give SMBs access to the technology without the capiital expenditure. Amazon, for example, offers Business One with HANA database and compute power on Amazon Web Servers for a pay-as-you-go hourly fee.
"Amazon offers small companies the choice of shifting computer tasks, computer intensive work out there to the cloud with very limited cost," he says. "That may allow retailers with the skills to do more exploratory work, or assist smaller specialised companies servicing larger ones."
"It's all about this democratisation of in-memory computing that is happening right now with the ability to purchase computing power and software in very small increments."
Gatner's Pezzini says that many SMBs are also already likely to be using IMC though subscriptions to software-as-a-service (SaaS) cloud services.
"Quite a lot of SaaS applications are based on in-memory computing types of architectures," he says. "It means many businesses won't know they're using it explicitly, but it's embedded into certain types of applications, such as logistics, planning and other things of this nature."
Pezzini gives the example of Workday, the cloud-based financial and human capital management soutions provider, which is built on an in-memory architecture.
"In-memory computing enables Workday to implement functionality and provide the scalability and performance that would be very hard to implement using traditional architectures," he says. "They don't market the fact that they use the technology or give users a choice of whether they want it or not. It's just there."
IDC's Lykkegaard adds that cloud-based social media monitoring services, which allow businesses of any size to instantly analyse Twitter and Facebook activity, are also driven using IMC technology.
"This is another way of how these small enterprises will start to use in-memory computing," he says. "It's just going to be increasingly under the hood of some of these services and software that they are already using."