NonStop for Dummies - A Summary
HPE NonStop is an integrated hardware and software solution that has been designed and built to survive the loss of components is considered to be critical to other systems.
It is a self-healing platform, with the ability to reconfigure itself in the case of failure. HPE NonStop, or Tandem, as it is also fondly remembered - is used by organizations who need their systems to be always on.
HPE NonStop today
HPE NonStop servers have been around for more than 50 years, and are the backbone of core systems that most of us don't think twice about using today.
Immigration, manufacturing, ATMs and point-of-sale systems (POS), are just some of the applications that depend on HPE NonStop to reduce global downtime to a couple of seconds of unplanned downtime a year.
Common Characteristics of NonStop Workloads
The exponential increase in data generated demands for mission-critical data to be have data integrity and availability. This means that the data generated has to be consistent and available at any point in time.
In the industry, operations on such data is known as a transaction - either the entire operation happens, or nothing happens. For example, either money is withdrawn from your savings account, and dispensed at the ATM you're getting it from, or nothing happens at all.
The system either performs all the steps required to complete the transaction, or none of them.
Fault Tolerance and High Availability
Aside from maintaining transaction and data integrity, fault tolerance and high availability have also been built into the NonStop. Fault tolerance in the NonStop mean that no single point of failure will impact the system.
High availabilty, as defined by the IDC (Intenational Data Corporation), means that an application is available even in the face of failure in part of the system, avoiding any planned or unplanned outages. This translates to a total downtime of about 30 seconds a year, at 99.9999% uptime.
Data Integrity and Reliability
Data integrity is the quality of maintaining data correctly, despite flaws in the surrounding systems, and a reliable system is one that would rather not return a result, rather than returning a wrong result or corrupting a database.
A mission critical system should be one that is reliable, available and fault-tolerant.
What makes the NonStop nonstop
In the market today, a variety of hardware platforms run an even more diverse range of operating systems, and aside from a few other vendors, most hardware and software pieces seldom come from the same vendor. HPE NonStop's tight hardware and software integration enables fault-tolerance and its self-healing capabilities to be consistent throughout its hardware and software components.
Clusters vs NonStop's Fault Tolerance
Quality servers with RAID (redundant array of independent disks) arrays is one of the ways to provide basic fault tolerance to storage. However this level of fault tolerance is often insufficient for most applications - hence the invention of clusters.
Clusters are a set of tightly coupled servers and systems sharing resources, to protect mission-critical workloads from certain failures. Clustering is a complex task - requiring meticulous load balancing and fine tuning, especially with syncronous data between individual systems.
NonStop's hardware reinforces redundancy to the extreme, providing effectively 100 percent uptime - every hardware component on the NonStop system is duplicated. This means that there is at least two storage units, two processors and two RAM busses.
Self-Healing Software
From a high-level overview, the HPE NonStop system can be described as a self-managing, self-healing cluster in one OS image. This is possible because of the shared nothing architecture, preventing malware and viruses from spreading between the components of the systems.
Critical operating system processes run in what is called process pairs in over two CPUs, allowing the processes to survive failures of any one particular CPU.
Because of the design, the software is able to transparently fail-over over to the second set of components without interrupting existing operations.