In an industrial environment and not only (these are issues initially born in the financial sector) the increase in the availability of systems up to Fault tolerance has become one of the most discussed topics.
Who sells hardware claims there is a need to Increasingly "armored" servers.
Who sells Software speaks of redundancy, sync and digital twin.
Who is right? Let's try to do some reasoning.
In the first instance, the "availability pyramid" must be defined in which the KPIs are the RTO and the RPO, i.e. the recovery time and admissible data loss (again in terms of time).
Fault Tolerance and High Availability are the two highest steps of the pyramid: we continue by defining the two ways to pursue the goal.
Software high availability
The high availability software it is the type of HA that many people are familiar with.
This refers to the systems uptime, typically obtained from replication software.
These software replicate the data on one or more different sites, so that when the primary system is offline their availability goes into "failing over" on the secondary site (which has a copy of the data), one of the critical factors, however, is the obsolescence of the copied data.
For example, if the data replication is not in real time, the work done since the last copy performed is lost until the system crashed.
Hardware high availability
The high availability of the hardware is the physical aspect of HA solutions.
This refers to having redundant systems, such as a secondary or even tertiary node ready for recovery when the primary system goes offline.
In order to keep your systems highly available, it is You need to have a secondary set of fault tolerant hardware to collect all the complete system.
An example of this is RAID (Redundant Array of Independent Disks), which many companies use to copy their data across multiple hard drives.
The difference between software HA and hardware HA
Companies often confuse high availability of hardware (i.e. secure, redundant, and fail-safe physical systems) with the high availability of the software.
HA hardware is a critical component of HA software and is not enough to keep systems available without good HA software as well.
To achieve system HA, you need effective replication software, error detection mechanisms, error recovery as well as consolidated network hardware redundancy, and real-time node-to-node disk synchronization, so as to restart in the same state in which the system crashed.
For a data logging system and archive a Cloud or remote HA is also possibleinstead for industrial applications, In which real-time usage is essential an internal HA system is recommended in order to remove latency due to the connection between the site and the remote system.
In conclusion…it is better to pursue HA with Software or with Hardware? obviously Can NOT be answered to the question thus posed.
First of all application considered critique must be handsome either per quanto riguarda physical redundancy (suitable pc and server) than the digital one: therefore hardware sufficiently "bodied" and a constant synchronization between (at least) 2 software sessions. This is to avoid any single point of failure.
In the industrial environment, the solutions that most look to the future are those that fall into the EDGE category: today there are industrial PCs capable of guaranteeing very high levels of availability and at the same time guaranteeing a secure bridge to the CLOUD, the ideal environment for storing storage and analysis tools.
An example? Click below!