By Greg Holmes
Converged systems are integrated datacenter systems which combine hardware and software together in set configurations that are designed for optimal performance and simplicity in terms of running a datacenter. As described by David Vellante of Wikibon, “Converged infrastructure (CI) is an integrated set of compute, storage, and networking components with infrastructure management software that provides a single logical chunk of hardware and software that is either specifically engineered together or at the very least tested and proven in a variety of configurations and applications. Converged infrastructure simplifies hardware and software management and accelerates the deployment of infrastructure for private clouds.”
If you look at the main vendors such as HP, IBM, Oracle, Dell, Cisco, EMC, NetApp, VMware, and VCE, whether it’s called converged systems, infrastructure, or engineered systems, there is an offering out there targeted at the kind of infrastructure your organization needs to run.
It is also an area the above vendors are pushing heavily. You can’t go through an airport now without seeing a 20m billboard touting how much faster Exadata makes your business run. The total available market for converged infrastructure is growing fast too; by 2017 it will represent about $402 billion, replacing datacenter spend that is predominantly legacy purchasing of hardware and software that is put together in house. (Figures from David Vellante of Wikibon)
Figure 1 - Illustrative cost difference of running on Converged Infrastructure vs Legacy systems
These systems bring together hardware resources such as CPU, memory, and storage – including fast storage like SSD, and cheaper storage like SAN and HDD arrays – and the software that organisations want to run, in tested and flexible configurations. The hardware itself is not necessarily cheaper; however it will perform better and require less support.
The advantages of converged systems are clear. InformationWeek has pointed to two main long-term advantages of a converged infrastructure. ("Why 'Unified' Is The Hot New Idea For Data Centers," George Crump, InformationWeek, March 14, 2009)
1. Lower costs as the result of both:
- Lower capital expenses resulting from higher utilization, less cabling, and fewer network connections
- Lower operating costs resulting from reduced labor via automated data center management and consolidating storage and network management infrastructure teams
2. Increased IT agility by:
- Virtualizing IP and Fiber Channel storage networking
- Allowing for single console management.
The main use case for this kind of system is to run mission critical apps and platforms including, but not limited to:
- Virtual System Infrastructure (e.g. VCE VBlocks running vCenter)
- Database systems (such as Oracle Engineered Systems & Databases – e.g. Oracle Exadata)
- Data Warehouse solutions (HP AppSystems running HP Business Data Warehouses & SAP)
- Cloud Infrastructure (involving storage, VM and Cloud automation)
The problem still remains, of course—how should you license the installed software? At first glance, a naïve observer might feel that since these systems come with all components installed, then all the required software licenses are included too. However, this is not usually the case and the licensing models for these products are often complex. Furthermore, many organisations may already have licenses to allocate for use in the datacenter. Apart from necessary OEM licenses (e.g. for Operating Systems) the licensing problem still exists, and is as complex as ever.
The licensing problem is exactly the same as when installing commercial software products on regular hardware put together in-house. You still need to count against the vendor’s standard metrics, measure usage, understand and apply your software product use rights, determine an accurate license position, and buy accordingly.
This means you need to be able to do the following:
- Collect hardware and software inventory, and software usage information
- Reconcile license consumption (installations and usage) against the license entitlements you have purchased. Some licenses might just concern the number or types of users you support – the processing power might be irrelevant!
- Calculate the number of licenses you really need – there may be better ways you could apply the licenses you own, based on your software product use rights, to reduce license consumption.
These converged systems usually run standard tools that you would install on your own machines. The advantage is that these tools become more mainstream because of their inclusion in these integrated systems. Converged systems will contain more comprehensive information that will be useful to sophisticated Software License Optimization tools.
The license types that organizations run on converged systems can be complex. There may be the right to have multiple instances under a single license or each instance might need to have its own license. At the end of the day, it comes down to detecting the instances and measuring the resources available to them, either directly or indirectly at the physical machine level. In some cases, it is possible to run an instance of an application on a tiny fraction of the whole machine’s capacity and just buy that amount of software entitlement (so called ‘sub-capacity’ licensing).
Sub-capacity licensing may involve multiple layers of virtualization. For example an Oracle SuperCluster allows for 4 logical domains (LDOMs) each running its own independent Solaris OS, with separated Zones within, and the potential to run Oracle virtual machines (VMs) within those, or have your application run with a hard allocation of processor power within an already restricted zone. Your application instance might have a single thread of processor power, or could access all of the processors allocated to the LDOM.
These complex instances and related applications have side effects on licensing, configuration management, and security compliance.
Virtual systems also mean you could have a variance in day-to-day use, as you spin up machines based on demand to meet your data processing needs. This means you will need to monitor peak usage in terms of the number of virtual machines running a particular application, to ensure enough licenses are available. The ability to track trends in usage and even undesired usage will help avoid unwanted spikes in demand and maintain a lower level of software cost. It could be that you need to set rules to restrict virtual machine movement and clustering to avoid unnecessary licenses and capacity. This is illustrated in the figure below where DRS Host Affinity Rules are applied to prevent VMs from being moved to particular servers.
Figure 2 - Dynamic virtual environment with VM move restrictions based on host affinity rules
In some cases you buy subscription based licenses which allow some variance according to how you use them. So there might be flexibility in licensing models which could help the organization manage their license consumption and costs tightly.
Being able to simulate the effect on these different license models and understanding whether your existing license entitlement can accommodate a migration to one of these converged systems is important. You must be able to plan your migration for not only the hardware, but the software as well; then you can budget for the change. You may find that because these systems run faster (10x faster according to some advertising) on the same number of processors, you get more performance, and therefore might be able to make do with fewer licenses. A good result, if you are spending money on new infrastructure.
Most organisations only replace their hardware as it is necessary – over the course of a 3-year or 5-year refresh cycle. Converged systems are typically sold on the agility, performance and ease of management benefits. Ensuring that an organization is able to measure the cost and plan for the ongoing use of the software that the hardware is running is a critical aspect of moving to a converged system. Organizations that can understand and optimize their license consumption and plan for change will reap the biggest benefits, without making costly mistakes in the process.
***
To learn more about Software License Optimization, please view our on-demand webinar: Software License Optimization Made Easy
or, read our white paper: What Does it Take to Achieve Software License Optimization?