PRIMEQUEST means growth without growth pains
Most businesses never stop growing. This means their infrastructures must also grow. The question is how to strike a balance between your bottom line and the costs of expanding systems including installation, operation, and maintenance charges?
This article looks at these costs and clearly shows how they can be reduced by choosing the right mission critical server.
Typically, there are two types of server growth. These are scale-out (multiple servers side by side) and scale-up systems (larger servers that can scale internally). You might think that using a scale-out system approach based on generic x86 servers would best help you reduce system life-cycle costs, especially compared with using a single and potentially larger mission critical server.
However, it is important to understand the hidden problems that occur in scale-out systems.
- Expansion in scale-out systems involves increasing operational costs
- The scalability of parallel clusters is limited by the application's parallelism
- Dedicated cluster software incurs additional charges
This article looks at how using a scale-up system approach using PRIMEQUEST mission critical servers can eliminate such hidden costs.
Scale-out server infrastructures create complicated installation and maintenance issues
Problem#1: Cluster system expansion can generate high costs
The more complex the system becomes, the more the system costs grow. For instance, all servers in a cluster system must have identical OS, firmware, and driver configurations. This is because even a subtle difference between servers may cause malfunctioning of the cluster. This means upgrading and application of patches must be executed to all the servers in the cluster.
Upgrading a cluster system requires processes below:
- Hardware health checks
- Hardware installation
- Addition of new resources to cluster system
- Operational tests
- Addition of resources for monitored objects
- Addition of resources for backup objects
- Update of operational manuals
Figure 1. Installation processes comparison
Between scale-out and scale-up servers
Problem#2 : Scalability of clustered system may be a constraint
The scalability of the system resources is an important discussion points when choosing a clustered system.
If you choose a parallel cluster system, limitations in application parallelism may hinder the scaling-out of the applications. In addition, because most of applications need to coordinate between multiple servers, you may have to implement application parallelism from scratch.
Oracle Real Applications Cluster (RAC) would be an option because it can coordinate between servers. But even in that case, you would still be required to modify the relevant applications and match them to Oracle RAC's application interface. Such application modifications are necessary to cancel failed SQL executions such as INSTERT or UPDATE.*1
*1 For further details see
Problem#3: Software licensing and support charges add costs
You will need to examine what additional costs you are required to pay for your cluster system. In addition to the cluster software products, you must also pay for OS and DB optional products.
- If parallel cluster DB is used, Oracle RAC is charged based on all the CPU cores in the cluster system. So, an Oracle RAC system is charged at 50% higher than a single server system with the same number of cores.*2
- Some OS are charged to all the servers in cluster system.
*2 Processor License of Oracle RAC is US$ 23,000 and that of Oracle Enterprise Edition is US$ 47,500. For the detail, refer to "Oracle technology Global Price List". http://www.oracle.com/us/corporate/pricing/technology-price-list-070617.pdf
PRIMEQUEST can reduce system life-cycle costs
PRIMEQUEST can help you grow your system while reducing operational and installation costs. PRIMEQUEST allows you to add system resources just by mounting additional system boards into the server chassis.
- PRIMEQUEST's mid-plane design enables easy plug-in
- This enables painless up-scaling of applications
- It also minimizes software cost growth
Business Benefits#1 : Simplest possible system expansion from PRIMEQUEST Mid-Plane design
Designed for scalability and flexibility, PRIMEQUEST can help you grow your system painlessly.
At the heart of the design is the Mid Plane. This acts as the hub for all components and eliminates complicated installations. When you open the cover of your PRIMEQUEST, the only wires you will find are power cords. Importantly, for ease and speed of maintenance, you can install components including System Boards and I/O Units while the chassis remains in its rack. This removes the significant time it would take to move heavy servers from rack to rack.
Figure 3. PRIMEQUEST Mid Plane
Business Benefits#2 : Painlessly Application Growth
You can expand the resources available to applications such as number of CPUs, memory, and I/O, without any modification to the application itself.
You can also optimize system resource use based on the performance monitoring tool provided in ServerView Operation Manager. This intuitive and well-integrated tool makes it easy to identify system resource shortages and the causes of performance slowdowns.
- Use of CPUs, memory, networks, and storage can all be recorded
- System resource shortages can create console alerts
- Performance graphs enable analysis of system performance problems over time.
Business Benefit#3 : Contained software costs
Scale-up servers like PRIMEQUEST do not require cluster software including Oracle RAC. So, single PRIMEQUEST is charged 33% less Oracle Database license than that of clustered system at the same number of CPU cores.
In addition, PRIMEQUEST hardware partitioning can shrink software license costs.