Infrastructure Management Solutions: Server Profiles and Templates
In This Article
In our first article in this infrastructure management (IM) series, A Look at Infrastructure Management Industry Leaders, emphasis was placed on what IM is and the need for IM in our datacentric world. It also outlined the IM solutions offered by three original equipment manufacturers (OEMs) -- Cisco, Dell Technologies (Dell) and HPE -- and explained how each solution came to market, as well as a general overview of their workings. This second article takes a deeper look at the concept of server profiles and templates for each of the three IM solutions.
Server profiles are logical groupings of settings for a particular server. In other words, these profiles are used to define the role or roles associated with a server. Specified settings can include BIOS settings, network interface card (NIC) settings, media access control (MAC) address settings, etc.
Correspondingly, server profile templates are more generic in nature and can be applied to multiple servers. Think of the server profile template as a cookie cutter carving out individual, identical profiles (the cookies) to apply to the server.
The beauty of the server template is that it is created once and applied many times to different servers, which controls configuration drift. For example, if there is a cluster of three servers and the same configuration needs to be applied to all three servers, a template would be utilized so that each server is set up exactly the same. Not only is each server in the cluster set up exactly the same, but the work of setting up the server configuration was only performed once and then simply applied to each of the other servers.
It is easy to see the benefits of using server profiles and templates from an IM perspective. With server profiles, servers can be managed from one central area. Server profile templates can make server configuration in a data center effortless by offering the ability to apply the same server settings/characteristics to several servers. This prevents redundant work, as well as the added chance for mistakes.
Overall, server profiles and templates make the management of an organization's infrastructure much more effective and efficient.
To further understand the concept of server profiles and templates, it is helpful to take a look at the history of profile management, as well as what the industry coined as stateless computing. Stateless computing and profiles came to the forefront when blades and converged infrastructure were released.
Stateless computing embraced three key components. One component is the idea of booting from a storage area network (boot from SAN). A second component of stateless computing is the virtualization of server parts previously located on physical adapters and physical servers, such as World Wide Port Names (WWPNs), World Wide Node Names (WWNNs), MAC addresses and universally unique identifiers (UUIDs). The third component of stateless computing involves policies, which instruct the server on how to behave.
What benefits did these developments bring to IM? To sum it up in one word: portability. These abilities made it easier to make adjustments to the physical aspects of servers. For example, if the motherboard went out on a server, the need for immediate replacement was unnecessary. Instead, the profile could just be reassigned to a different server with a working motherboard, resulting in little to no downtime.
In the past five years, all three OEMs have gone beyond the original concept of server profiles and templates. Instead of these profiles being applied just to blades/converged infrastructure, server profiles/templates are assigned to rack mounts, storage systems, switches, etc. Additionally, the OEMs have allowed their systems to be used in conjunction with automation tools such as Ansible, Chef, PowerShell, Puppet and more. This enables organizations to use their preferred automation tool(s) when creating profiles.
Now that there is an understanding of the history of server profiles and templates, as well as a general overview of their proficiencies in recent years, let's look specifically at how each OEM utilizes server profiles/templates in addressing IM.
Cisco initiated the server profile concept with the release of its UCS Manager in 2009. While HPE's C7000 already had virtual connect, giving its users the ability to assign virtual connect profiles, Cisco's UCS Manager was able to take this same idea and apply it to everything inside of a Cisco UCS server. Cisco's IM solution is considered entirely stateless.
Within UCS Manager, a master template is configured and comprised of pools, policies and templates. All settings, updates, etc. are done in Cisco's software. Once the template is created, service profiles can be deployed to the servers. If a change is made to the master template, it propagates to the service profile. This is known as a parent-child relationship, wherein the parent is the master template and the child is the service profile. This concept offers very fast and efficient server management especially in environments where multiple servers are being managed.
Cisco does not have the same problem that its competitors do as far as maintaining support for older versions. With Cisco coming to the market in 2009, Cisco is just now starting to have "legacy" equipment. Consequently, Cisco UCS Manager works with most generations of servers (i.e. Gen3 and newer).
Cisco expects to release its newest IM tool, Cisco Intersight, in the near future. Cisco Intersight will do everything that UCS Manager does but it will be in the cloud for ease of use anywhere, at anytime. Intersight will monitor all systems and allow customers to use the Connected-TAC (call home) feature without purchasing any licensing. Management in Intersight, though, will require the purchase of an essential license.
HPE initially dabbled in server profiles with their product, Virtual Connect on the C7000. Virtual Connect had some sense of profiles, but only presented the ability to manage the network and SAN connections of a blade server. In order to encompass full server management capabilities, HPE introduced OneView. In doing so, HPE offered a profile management system inside of a Linux-based virtual machine (OVA), allowing for more comprehensive server management in the data center.
The list of settings that can be defined in OneView is extensive, including the specification of settings for RAID groups, UUIDs, host bus adapters (HBAs), BIOS, etc. Basically, all settings that can be applied manually to a server can be applied through the server profile.
HPE has a distinct way of assigning Profiles to hardware generations/hardware types. This means that with each new hardware release, a new drop-down menu will correspondingly be assigned to that particular hardware type. This helps control configuration drift across multiple hardware types. HPE only allows for the selection of hardware types that are currently assigned to the OneView system, preventing potential accidents and mistakes.
There is a limitation in OneView that should be mentioned. OneView is not capable of managing Gen7 servers or older. Not all customers see this as a constraint, though, as most organizations likely will not want to run production workloads on outdated hardware.
OneView offers a fee benefit; it can monitor many systems without the need for licensing. However, to actually use the Profile Management feature of OneView, an ILO Advance License is needed.
Dell takes a slightly different approach to the idea of server profiles and templates than its competitors. Dell's IM solution is OpenManage Enterprise. As noted above, the creation of server profiles and templates in the IM solutions offered by Cisco and HPE is done through each of their respective management software. As such, Cisco UCS Manager and HPE OneView employ a push mentality, wherein everything is set up on the management console and then pushed down to the blade/rack mount.
Dell, on the other hand, does not utilize its management software as the first step. Instead, Dell allows for the setup of servers outside of or without the use of management software. Once the server's settings are established (i.e. RAID, BIOS settings, etc.), the profile can be pulled in through OpenManage Enterprise as a template to use for other servers.
One may be asking the question: "How can the template be updated and pushed out to the blades and rack mounts when there is just a network change?" A separate template is needed to address each network change. For example, a template for NIC configurations is needed in order to add VLANs to trunk ports. In this case, the template is strictly centered around the NIC configuration and nothing else. Once this template is established, the network where the VLANS are to be added as a trunk can be edited to accommodate this addition.
It is important to note that care should be exercised when invoking a destructive event such as reconfiguring a RAID group. The whole server could potentially be wiped clean by doing this.
Our subject matter experts are here to help you navigate through the top IM solutions on the market today. Using our state-of-the-art Advanced Technology Center (ATC), our IM and ATC experts can further explain these concepts and offer you the unique opportunity of trying out the various IM platforms.