By Jason Andersen, Vice President – Business Line Management, Stratus Technologies
The use of software as a service (SaaS) and infrastructure as a service (IaaS) is well established in the enterprise data center. Now, industrial organisations are beginning to make tentative steps to XaaS offerings for their operational computing. However, there is some trepidation preventing many from taking the leap. How will it impact the way they work? How will it change their budgeting process? In the operational technology (OT) world, change is not always viewed as a good thing. For those industrial organisations looking at XaaS, there are some important considerations.
For starters, the OT infrastructure may not be ready for XaaS. Over in the data center, connectivity and robust security are generally well established, so linking up with SaaS offerings is pretty straightforward. But in the OT world, this may not be the case. Upgrading the network infrastructure to enable secure, high-performance connectivity to XaaS services may involve installing gateways and/or specialised software, potentially representing a substantial investment of time and money.
Another consideration is the ongoing investment required for XaaS. A common example is an organisation collecting data at the operational edge looking to push it up to cloud to perform analytics. However, getting the most from such analytics requires specialised skill sets that many OT organisations do not have. Moreover, depending on the size of their OT infrastructure, the volume of data generated can be enormous, requiring lots of costly “leased” capacity.
So when does XaaS makes sense for industrial organisations? In selecting workloads for uploading to the cloud, consider the following:
Applications vs. infrastructure
SaaS offerings that offer a pre-packaged, turnkey solution for a particular application may be the best fit for OT. Offloading your supervisory control and data acquisition (SCADA) system or Historian can be a good way to leverage the cloud, with minimal impact on your budget and available skillset. However, be aware that many of these services may require you to do some work at your edge, such as adding a gateway, which adds cost and complexity to the move.
Applications with low support requirements may be the best candidates for SaaS implementations. That’s because SaaS is inherently “low IT touch.” Applications that have higher support requirements are probably better remaining on premise, especially for organisations that already have solid IT organisations backing up operations.
Data volume and criticality
As previously noted, cloud computing, memory and I/O capacity can get expensive very quickly. Outsourcing applications that handle large amounts of data may be cost-prohibitive. For example, a SCADA system measuring more than 1000 “tags” might be more cost-effective to keep on premise. In addition, for applications that handle critical data requiring high levels of availability, keeping these in-house on fault-tolerant systems may be prudent.
Remember that the XaaS question is not binary. Making thoughtful choices about what can be outsourced and what is best kept in-house may well lead to a hybrid model—with some computing happening at the edge and some in the cloud—that offers the best of both worlds.