We heard from another customer today that their incumbent cloud security vendor keeps going down. And when it is not down, DLP scans take hours, if they complete at all. What is going on?
Cloud security offerings are typically based on the network security model, where fixed capacity nodes handle traffic on fixed bandwidth pipes. These "bump in the wire" offerings are priced and sold by bandwidth, and things work well enough when they are used in such single-tenant contexts.
But modern cloud security services are sold by user, and must handle usage on any app, any device, at any location. "Bump in the wire" network security architectures cannot deliver on this model since users may move around and usage may fluctuate greatly. For example, a company could have an offsite Sales Kick Off, creating a surge at a remote data center. Or a company wide email with lots of attachments could flood mailboxes across thousands of users. In each such situation, the network security architecture is brought to its knees. DLP scans fail, and the system goes down.
Enter the Bitglass Polyscale Architecture. In this model, each data center starts with a baseline configuration of components comprised of databases, elastic search nodes, Hadoop nodes, DLP nodes, Malware engines etc. Each node communicates in a secure and stateless fashion with other nodes. If the load on any node crosses a threshold for a sustained period, the node automatically clones itself.
During the workday, each data centers swells and ebbs as loads come and go. When a customer has a major offsite, the remote data center expands towards the load profile of that customer. In short, the Polyscale architecture enables each data center to automatically morph during the day to handle workloads.
The key benefit of the Bitglass Polyscale architecture is that performance and uptime are stable regardless of load.