It was back in 2004 when SAP’s Irfan Khan held a press brief­ing to explain how mobile appli­ca­tions should be built. These days, Khan is GM and Glob­al Head of SAP’s Data­base and Data Man­age­ment divi­sion, but back in 2004 he was Chief Tech­nol­o­gy Offi­cer for Sybase. SAP famous­ly acquired Sybase for its mobile capa­bil­i­ties, and the rest is history. 

“Devel­op­ers should build mobile appli­ca­tions that obvi­ous­ly take advan­tage of con­nec­tiv­i­ty because that’s what makes them mobile — but it’s impor­tant to always design and build as if the con­nec­tiv­i­ty pipe is not there,” said Khan. 

Pre­scient Proph­esy of Life on the Edge

Khan’s prophet­ic words were not relat­ed to the Inter­net of Things (IoT), because the term didn’t enter our tech­nol­o­gy lex­i­con until much lat­er. But his point was well made: Access to the net­work isn’t always going to be an option because con­nec­tiv­i­ty drops some­times and some loca­tions (such as under­ground con­crete bunkers, oil rigs, air­craft, etc.) may rarely, if ever, be in range. 

Since Khan’s pre­scient mus­ings and the birth of cloud com­put­ing a few years lat­er, we have cre­at­ed a world where intel­li­gent con­nect­ed devices log in to net­works and pro­vide us with data ana­lyt­ics from a cloudy back end. While we may have solved some chal­lenges with cloud com­put­ing, we’ve cre­at­ed anoth­er: The back end is a) not always avail­able and b) some­times takes too long to com­mu­ni­cate with us if we want real-time per­for­mance, even if there is a con­nec­tion. Wel­come to edge computing.

What We Mean by Edge Computing

Edge com­put­ing (and edge ser­vices) describes the abil­i­ty for devices to per­form a degree of com­pu­ta­tion­al pow­er local­ly. To be clear­er, when we say local­ly, we mean on the device itself and with­in a defined perime­ter of oth­er net­worked devices and sen­sors. In tech­ni­cal terms, you could call this dis­trib­uted peer-to-peer ad hoc net­work­ing. In a layperson’s terms, you could call this a group of devices, sen­sors, and machines in close prox­im­i­ty with enough intel­li­gence to be able to talk to each other.

How close? We’re glad you asked. Ana­lyst firm IDC defined edge com­put­ing as a, ​“mesh net­work of micro data cen­ters that process or store crit­i­cal data local­ly and push all received data to a cen­tral­ized data cen­ter or cloud stor­age repos­i­to­ry, in a foot­print of less than 100 square feet.” 

The Laten­cy Factor

In real terms, edge com­put­ing often comes into play in finan­cial ser­vices or man­u­fac­tur­ing sce­nar­ios where device-to-data cen­ter (and back again) laten­cy means a com­put­ing sys­tem will fail, even when that laten­cy fac­tor is only mil­lisec­onds. Why all this clar­i­fi­ca­tion? Because SAP is adding SAP Edge Ser­vices for con­nect­ed assets and smart key per­for­mance indi­ca­tors (KPIs).

Dur­ing its appear­ance at Inter­net of Things World in San­ta Clara this month, SAP explained how cus­tomers can use SAP Edge Ser­vices with the SAP Asset Intel­li­gent Net­work to over­come work­flow dis­rup­tions due to inter­mit­tent or slow con­nec­tiv­i­ty. The SAP Asset Intel­li­gence Net­work is designed to bring machine oper­a­tors, engi­neer­ing ser­vices providers, and man­u­fac­tur­ers togeth­er into a col­lab­o­ra­tive cloud-based group to opti­mize the per­for­mance of oper­a­tional equip­ment. It’s what SAP calls ​“unin­ter­rupt­ed pro­cess­ing in dis­con­nect­ed high-laten­cy sce­nar­ios.” Or, if you pre­fer, it’s edge computing. 

Flex­i­bil­i­ty to be On or Off

Mala Anand, Pres­i­dent of SAP Leonar­do and Ana­lyt­ics, has point­ed to this software’s suit­abil­i­ty for dis­con­nect­ed envi­ron­ments such as off­shore oil rigs, min­ing, aero­nau­tics, and defense deploy­ments. She explains that SAP Edge Ser­vices allows cus­tomers to replen­ish mate­ri­als and sup­plies, exe­cute main­te­nance work, main­tain project vis­i­bil­i­ty, and per­form oth­er func­tions even when offline — and are still able to syn­chro­nize with their core sys­tems when con­nec­tiv­i­ty is restored. 

Par­adise by the Dashboard

SAP has also pro­vid­ed what it claims are easy-to-con­fig­ure dash­boards for users to design edge ser­vices man­age­ment por­tals. Accord­ing to this technology’s prod­uct release notes, the SAP Smart Busi­ness cock­pit helps teams visu­al­ize and draw insights from key per­for­mance indi­ca­tors (KPIs). It pro­vides ana­lyt­i­cal drill-down views to com­ple­ment the SAP IoT Appli­ca­tion Enable­ment toolk­it, with­out requir­ing addi­tion­al coding.

“Using appli­ca­tion tem­plates, IoT devel­op­ers can con­fig­ure and embed visu­al­iza­tions into IoT appli­ca­tions by drag­ging and drop­ping them on the page. In addi­tion to pro­vid­ing visu­al­ized per­for­mance insights, SAP Smart Busi­ness helps teams to man­age by excep­tion, with alerts when KPI thresh­olds are breached,” said SAP in a brief­ing statement.

Too Close to the Edge?

So, edge com­put­ing is an impor­tant part of the Inter­net of Things. Peo­ple also use ​“fog com­put­ing” to describe the notion of the wider net­work activ­i­ties that also occur near edge com­put­ing devices. But we won’t go down that road today. 

Speak­ing to ASUG in May 2018 to update his com­ments on this sub­ject, SAP’s Khan has explained that, ​“When design­ing mobile apps for edge-based pro­cess­ing, a fun­da­men­tal design prin­ci­ple is to incor­po­rate ​‘occa­sion­al­ly con­nect­ed’ per­sis­tence for the even­tu­al­i­ty of poor cov­er­age. Regard­less of the hyper­con­nec­tiv­i­ty avail­able today, black spots are still res­i­dent in both cel­lu­lar and Wi-Fi net­works. Offline [data] per­sis­tence is what makes pro­cess­ing at the edge viable. For­mer Sybase tech­nolo­gies like SQL-Any­where are wide­ly used even today to pro­vide offline, but full, rela­tion­al data­base capa­bil­i­ties on a very small mem­o­ry foot­print, per­mit­ting both com­pute and stor­age independence.”

The Jury is Still Out on Security

Some prac­ti­tion­ers argue that edge com­put­ing is more secure because less net­work traf­fic must ping back and forth across the con­nec­tion pipe. Some prac­ti­tion­ers argue that edge com­put­ing is less secure because edge devices are inher­ent­ly small­er, low­er pow­er, less smart and there­fore more eas­i­ly pen­e­trat­ed. Which side of the coin is right? It depends.

SAP’s moves into the IoT space con­tin­ues as the com­pa­ny con­tin­ues to make sub­stan­tial invest­ments in this area, so it is no sur­prise to see them sup­port­ing edge com­put­ing in this way. It turns out the that saga­cious­ly affa­ble Mr. Khan was right after all.

Join one of our SAP Leonar­do web­casts to hear from Mala Anand, Pres­i­dent of SAP Leonar­do and Ana­lyt­ics and oth­ers as you learn how you can imple­ment your own IoT projects.