Local IT infrastructure
An often cited reason to move to a cloud model is the reduction in local IT infrastructure and support costs. Further, the cloud service provider accepts the responsibility of ensuring that hardware fails over in the event of a problem and resources are adequately load-balanced to deal with surges in required capacity.
While it’s often true that up-front costs of an on premise solution are higher, it’s not always the case as many organizations already have infrastructure in place. The DTA Middleware Engine is a lightweight, highly performant web application with a small footprint without expensive hardware requirements, and is designed to be deployed on existing infrastructure.
Redundancy can be handled through failover to another host (and the cloud explained below). To simplify this, the Total Alert solution is readily deployable as a virtual machine. Using a virtual machine management solution such as VMware HA, failover can occur to secondary virtual machines on the same physical hardware, as well as to virtual machines on other physical hardware. Failover is handled automatically without user intervention. The solution even avoids a single point of failover at the management level, as it can be deployed with up to five primary hosts in a single cluster. Each primary host has the capability of acting as the autonomous managing agent if the active primary host fails. Further, if a failover operation itself fails for any reason, for example a new host cannot be brought online due to a resource constraint, the Distributed Resource Scheduler automatically attempts to adjust the cluster, for example by bringing other hosts out of standby mode. The Distributed Resource Scheduler can also add standby hosts to the cluster in the event of surges in capacity.
In summary, the on premise model has the promise of offering greater operational control, flexibility and reliability than an out-sourced cloud solution. For this reason, we see the DTA EMNS distributed architecture as being a preferable delivery model for such a critical application.
Quality of Service
- Quality of Service (QoS) is an area where on premise has a major advantage over cloud services. Most cloud services offer Service Level Agreements (SLAs) based only on best effort, and in particular QoS metrics are often not defined at all.
- Latency – a cloud service, even if connected to the organization with an expensive dedicated fat pipe, is going to suffer from higher latency than an on premise solution simply because of the round-trip involved. i.e. Slower Alert Delivery and Alert Receipt. Further a cloud service using shared resources could introduce further latency if an alert is held in a queue due to a busy resource. When lives are at stake, seconds count.
- Performance – a local alert, raised locally, and intended to be delivered locally to multiple resources, is clearly going to benefit from a local infrastructure. When delivery time of critical notifications is of primary importance, sending an alert request to a cloud service, for it only to be sent back to the same location for mass delivery is obviously less than optimal. An on premise solution offers the fastest response times. When lives are at stake, seconds count.
- Throughput – an on premise solution can be designed to meet the capacity needs of the organisation. A cloud service would have to provide (at cost) dedicated resources to meet exacting throughput targets, which negates the cloud’s advantage of lower costs through the use of shared resources.
- Availability – we’ve seen in the previous section that an on premise solution can be highly available through the use of autonomous failover and load-balancing. Another advantage over cloud solutions is that another potential point of failure is eliminated, that is the internet connection itself. Even with a redundant internet connection, a cloud based EMNS is potentially unavailable in an INFOCON 4 situation where internet access might be severely restricted.
- Cloud as a HQ Management and Redundancy Option – an on premise solution can be configured to utilize the cloud so that an organizations HQ locations may always maintain a hybrid real-time bi-directional snapshot of all of its on premise DTA MNS deployments across all regions. As such, the functionality of a centrally hosted cloud MNS is not deprecated by a distributed architecture; it is actually enhanced! Because over 95% of all system overhead and bandwidth is managed by the local sites the bandwidth utilization in and for the cloud is miniscule, resulting in faster communications across the enterprise coupled with an exponential reduction in cloud costs. In the extreme circumstance where several layers of redundancy are unavailable locally, as a last and rarely used option the cloud can jump in and provide temporary MNS functionality until the local system comes back on line.
In summary, the on premise model offers better quality of service than an out-sourced cloud solution and for this reason it’s seen as being preferable for a performance critical application such as an EMNS system. The DTA EMNS has been designed from the ground up with high performance as a number one priority.