5 Traits of the Best Servers for Analytics
Business Record Staff Jul 13, 2017 | 11:00 am
5 min read time1,133 wordsBusiness Insights Blog, The Internet of Things
BY DENNY FISHER, Chief strategist, ACS
You want a car that lets you know when there’s a problem, a mechanic or dealership who responds quickly, and the ability to get back on the road as soon as possible.
It’s no different when it comes to processing your data. The Internet of Things has opened the floodgates to the type of data that can be collected from a larger variety of machines, appliances and devices. The ability to process large volumes of data is of the utmost importance.
A quality server can optimize your analytic tools enabling them to sort through and process millions of pieces of data to determine what is most critical for your business and therefore help you make better business decisions.
Here are five things to consider when selecting a server:
Select a server that consistently does its job – meaning your data is available and accessible any time you need it – and provides ease of diagnostics if there is an error, says Michael Belle, the Power Systems General Manager for ACS.
The server should continuously conduct internal scans for hardware and software problems. You want a server that will immediately report any issue directly to the manufacturer, which will then notify the client to replace any defective hardware. If the issue is part of the operating system, choose a server that notifies the manufacturer, downloads and installs the software patch or update, automatically fixing the problem.
A reliable server is low maintenance with built-in redundancy capabilities and parts that are “hot swappable,” so they can be switched or replaced without having to bring down or interrupt the system, says Elliot Smorodinsky, the Power Systems Technical Team Lead for ACS.
Remember our car analogy? If the server features described above were attributed to a Ford Explorer, repairs would unfold as follows. When there’s an issue with your vehicle, an indicator light would come on. This action would also notify Ford that a malfunction has occurred. Ford would then automatically notify the local dealership, who may send a mechanic with the necessary parts or tools to your location. For your part to be “hot swappable” it would be like a mechanic fixing your cruise control module as you’re driving down the interstate at 65mph rather than having to bring the vehicle into the dealership for repairs.
- Reliable Virtualization
An effective server offers users the ability to have more than one operating system running on one physical server. You want a server that has reliable virtualization built into the hypervisor/firmware rather than being a software add-on.
“Instead of three different machines, you can run all three on one piece of hardware,” Belle says.
This allows for increased scalability and greater efficiency.
A server with built-in security is vital, considering the range of IoT devices with little or no individual security of their own that run data through the operating system. You’ll want to make sure your data is protected and cannot be compromised by hackers or cyberthreats.
Choose a server that you feel confident has a high level of trust to safeguard data, including information that is proprietary or confidentially protected through laws such as the Health Insurance Portability and Accountability Act (HIPAA).
Built-in encryption and security monitoring are important components to the server’s security system. These steps monitor programs at the binary level and ensure they have not been altered by a cyberthreat and are authentic.
- Flexibility to Scale
As more data comes in through the IoT, there are more demands on the computing requirements. Calculations need to occur at faster speeds, and data needs to be immediately available. A server that has flexibility to scale can accomplish this with real-time monitoring and real-time data.
Businesses also need the ability to “scale up” if operations need to be combined into a single, more powerful server where everything runs on a single system, or “scale out” if operations need to be distributed among multiple smaller systems.
It’s similar to having all of your offices in a centralized location where there is economy of scale and reduced overhead costs, or spreading them out among multiple branch offices that can allow for redundancy to avoid the risk of a single-point failure.
Choosing between the scale up and scale out approach will affect your total cost of acquisition. Instead of one larger expense, scaling out can break down the total cost into smaller amounts.
You want a server that can support multiple units of work that need to operate at the same time, or can pre-schedule the work so it can be completed in an organized manner.
When you’re building a house, you wouldn’t hire one construction worker. You want multiple workers to work in tandem in order to complete the job faster. Having a server with no or low multithreading slows down the process in a similar manner to hiring a single construction worker, while a server with high multithreading allows for the work to be done faster and more efficiently.
Choosing your server
IBM Power is a server that fits all of these characteristics and is one of the best at handling data analytics. It was built with the newest technology to maximize its computing power and operate efficiently.
It is 99.999 percent reliable with 89 percent of its servers having less than five minutes of unplanned downtime across all systems worldwide per year. Unlike other servers its reliable virtualization is built into the hypervisor/firmware rather than being a software add-on and allows for up to 1,000 individual systems to do different tasks all on the same server. In addition, the virtualization hypervisor on IBM Power has never been successfully hacked, which makes choosing IBM Power a wise choice.
“You get more bang for your buck,” Smorodinsky says.
It’s almost twice as efficient as other processors before its multithreading capabilities are taken into account – think two construction workers versus one. With multithreading factored in, 16 workers are on the job compared to two.
IBM Power has long been the most reliable and best equipped server for analytics on the market.
You know you have volumes of data; that it needs to be sorted, analyzed and organized into a useful form; and that it needs to be protected from cyberthreats. Your next question may be: “How do I get all of this done?”
A third-party assessment can assist with an evaluation of your data needs and help you implement a strategy that will allow you to increase your return on your investment, make better decisions and alleviate your stress in trying to manage all of these initiatives.
Read our next blog to discover the value of third party assessments and utilizing Managed Services to help manage the challenges of IoT.
|Denny Fisher, Chief Strategist