The business of business is continuing to move out of the traditional datacenter. As we have talked about often here at The Next Platform over the past several months, data is increasingly being generated at the edge of the network by the rapidly growing numbers of connected devices. So systems and workloads that once existed primarily in core datacenters are now running more in the cloud, in branch offices, in gateways, and in other edge systems.
The enterprise IT environment is highly distributed and will continue to evolve in that direction, with some estimates being that 75 percent of all business data within the next few years will be generated at the edge, and given the growing demand for real-time processing and analytics of the data, more of that work will have to be done out there.
The rise of the Internet of Things (IoT) – which could comprise as many as 20 billion connected devices within the next couple of years – and the trend toward increased mobility has fueled the increasingly distributed nature of computing and emerging technologies like artificial intelligence (AI), machine learning and data analytics has driven a greater demand for intelligent compute resources outside of the datacenter. The need for real-time data processing and analytics and the growing costs in money and bandwidth in sending data back and forth between remote locations and the datacenter or public cloud also is a factor.
Not surprisingly, established IT vendors and startups alike are shifting much of their efforts away from the datacenter and out to remote areas like branch offices and the edge, essentially – along with the cloud – building a third environment for enterprise computing and creating an area that is quickly becoming a major area of competition. For example, for several years Dell EMC has built Intel-powered gateways that can aggregate and analyze data at the edge and then send useful data to the cloud or datacenter. The company also sells its dense PowerEdge R430 and R640 systems into edge environments and earlier this year introduced its Virtual Edge Platform 4600 to bring its open networking concepts to the edge. VMware has expanded its NSX software-defined networking (SDN) from a product to a portfolio of products that touches not only the datacenter but also hybrid cloud environments and branch offices. Cisco Systems continues to push its hardware, software and services beyond the datacenter and into the edge and multicloud environments.
Similarly, Hewlett Packard Enterprise is building out its portfolio for what it calls “the intelligent edge,” which saw revenues in the most recent quarter jumped 17 percent. In a call with Wall Street analysts last month, CEO Antonio Neri said that “customers tell us they want to take advantage of exploding amount of the useful data being created at the edge.” Aruba Networks, bought by the company three years ago for $3 billion, has become a key ingredient in HPE’s remote office and edge computing efforts, with the vendor moving its Edgeline converged systems and IoT gateways from its datacenter unit to Aruba.
The company now is enabling organizations to run unmodified enterprise workloads at the edge. At the company’s Discover 2018 show this week, HPE unveiled new solutions based on its rugged Edgeline EL1000 and EL4000 converged systems (below), which have the compute and storage capabilities to run enterprise-class software, including Microsoft’s SQL Server and Azure offerings, Citrix’s XenApp and XenDesktop, SAP HANA, ThingWorx from PTC, GE Digital’s Predix IoT platform and SparkPredict, AI-based software from startup SparkCognition. Both systems employ Intel Xeon D processors with eight to sixteen cores or quad-core Xeon E3 chips with a GPU accelerator. A key to the new solutions is the Edgeline Extended Storage Adapter option kit, which enables businesses to add up to 48 TB for software-defined storage to the Edgeline systems, giving it them the ability to handle such workloads as AI-based applications, databases and video analytics at the edge.
Most edge systems don’t have the compute or storage power to run full enterprise applications, Tom Bradicich, vice president and general manager of HPE’s IoT and Converged Edge Systems unit, said during a presentation at the show. (Bradichich used to be the architect of IBM’s System x server line and also helped Big Blue spearhead the creation of the ThinkPad notebook and the InfiniBand interconnect.) Instead they are forced to run applications – think Microsoft Azure IoT Edge – that are adapted to the capabilities of the systems. Through the Edgeline solutions, HPE is giving enterprises a third place to run the same unmodified applications that they use in their datacenters or the cloud.
“The cloud is the off-prem environment,” Bradicich said. “The edge is the other on-prem. It is the same high-performance, high-powered technology that we see in the cloud, but out at the edge.”
The solutions will enable enterprises to process and analyze the data at the edge, which will mean faster response times – which are crucial when thinking about such uses cases as autonomous vehicles – and reduce the bandwidth costs and latency that come with having to move data from the edge out to the cloud or into the datacenter. In addition, the applications at the edge can be managed by the same software tools used in the datacenter and cloud, whether from HPE or a third-party, and common security measures can be used. The systems also are ruggedized, which means organizations can run their enterprises apps in such difficult environments as manufacturing floors or petroleum or natural gas platforms in places like the Gulf of Mexico.
In a similar play, Aruba introduced the software-defined branch (SD-Branch), an integration of the various parts of a branch office network, like access points, switches, gateways, management, and software-defined WAN (SD-WAN) – into an on-premises appliance that can address the challenges from mobile apps, the IoT and the cloud. It can manage both Internet and datacenter traffic and enable enterprises to many more remote locations with fewer people. Kishore Seshadri, vice president and general manager of cloud solutions at Aruba, said during that show that on average, enterprises have eight to ten people managing 2,500 to 3,000 locations, a ratio that makes it difficult for them to scale their operations to meet the demands of increasingly distributed and disaggregated environments.