Ubuntu 22.04
(as of January 2023)
Ubuntu 24.04
(as of May 2024)
A typical Unix Managed Server deployment might consist of the following steps:
We manage the server and core features but customers have some responsibilities when using a Unix Managed Server
Routine requests can be submitted via email to help@uw.edu. Be sure to include the name of your managed server in the body of your request so that it will be routed to the appropriate team. Typically these queries will be responded to within 2 business days or less.
Emergency requests such as server-down conditions can be submitted to the UW-IT Computer Operations staff 24/7. Email can be sent to copstaff@uw.edu or phoned in to 206-685-1270. Be prepared to provide the name of the server and your contact information. Operations staff will page the on-call engineer for your system and the issue will be handled as soon as possible.
Typically, a server can be provisioned in four to six weeks from the point that resources are available. For virtual machines, there is usually no added delay. However, physical systems require roughly an additional two to three weeks in order to acquire hardware from the vendor.
While this is technically feasible, it may not make financial sense for the customer, and there may be better solutions to consider. For example, if you require physical machines to run in a UW-managed and secured datacenter, you may wish to consider the UW-IT Co-Lo service. Or, if you need a simple virtual machine, you could consider third-party hosting providers such as Linode or Rackspace, or even cloud providers such as Amazon's EC2.
Unix Managed Servers are only supportable for a limited amount of time. Availability of vendor server operating system support and security updates is the primary driver of the support lifetime. Upgrades to new operating systems are not automatic, you must initiate a new server request and migrate your services from your old server to the new.
| Operating System | Released | Standard Support Ends | Extended Support Ends |
|---|---|---|---|
| Ubuntu 24.04 | 2024-04 | 2029-04 | 2036-04 |
| Ubuntu 22.04 | 2022-04 | 2027-04 | 2032-04 |
| Ubuntu 20.04 | 2020-04 | 2025-04 | 2030-04 |
| CentOS 7 | 2014-06-10 | 2024-06-30 | 2028-06-30 |
| CentOS 6 | 2010-11-10 | 2020-11-30 | 2024-06-30 |
Extended support begins when Standard Support ends and is an additional cost that will be added to the cost of your Managed Server.
Virtual servers are most common and do not wear out but physical servers must also be replaced at least every 7 years in order to remain under hardware maintenance.
Read the UMS Operating System Upgrade Guide <link to KBA here> for instructions on how to upgrade your OS.
Ubuntu Server is the current supported operating system. We currently offer Ubuntu 24.04 by default or Ubuntu 22.04 if needed. If another operating system is desired, please contact help@uw.edu to inquire about a time and materials consulting request.
Unix Managed Servers are automatically monitored 24/7 throughout their life. Basic functionality (ping, administrative daemons) is checked, as well as many UWIT-managed applications. If a problem is detected, an on-call engineer is paged and the issue is handled as soon as possible.
Monitoring includes Prometheus metrics gathering. You can view a dashboard of various metrics, including CPU, Memory, Disk and Network usage here: https://graphs.s.uw.edu/grafana/goto/zEKOBFwIk?orgId=1
In most situations, we can arrange the system to allow you to do your work without needing root access, especially if we are permitted to manage the applications for you. However, it is ultimately your box, so if you want root access, we will give it to you. We simply ask that you not interfere with our management processes, which typically means avoiding alterations to system areas (e.g., /etc, /var, etc.)
Whenever possible, we use UW NetIDs to authenticate and permit access to Unix Managed Servers. At deployment time, we obtain a list of users from the customer who are to be permitted to log into the server. These users will then be added to a group that will allow them to enable access to the system via the UW NetID Management portal.
The cost varies from server to server and depends on the customers' individual requirements. If you would like to see some examples of past managed server agreements, check them out here.
We typically manage at least the hardware and operating system of the host. We can also manage many common server applications (see other FAQ entries).
A non-exhaustive list of applications UW-IT can manage on your server on your behalf includes:
Apache web server
MySQL database
Ubuntu systems are managed using Ansible. Please see Unix Managed Server Policies#SystemManagement
Our older systems (CentOS 6 and CentOS 7) are managed using an in-house configuration management system known as "ref." Unlike many commonly used management frameworks available today, ref is "strongly managed" in that it maintains very strict control over files it believes it manages. Ref maintains a reference copy (thus where the name came from) for every file it manages in the operating system and for each managed application. Periodically, ref scans every file on every managed system and if a change is detected, it gets updated to what ref thinks it should be.
From a customer's standpoint, this means that any change made in a protected area, such as altering a configuration file in /etc, will be silently and automatically reverted. Therefore, it is important to coordinate any necessary changes to the operating system or managed applications with UW-IT staff.
On a basic Unix Managed Server, we keep the operating system (Ubuntu/CentOS Linux) up to date by applying software patches and updates roughly quarterly. We occasionally release an isolated patch immediately if it fixes vulnerability that we deem to be high-risk.
If desired, we can also manage and update certain application and server software for you (such as web servers and databases); this is negotiated upon request.
We verify that nothing obvious will be broken by an update, but we can't exhaustively test every function you might use, nor can we test customer software.
We do, however, apply patches to a subset of servers a day in advance of releasing them to all systems. Many of our customers who have redundant or development servers in their cluster will choose one of them, called a "canary", to be on this list of servers which get patches ahead of the others, in order to verify functionality. Let us know if you have that requirement.
Typically, a Unix Managed Server will be provisioned with a /data filesystem. This filesystem is "excepted" from ref inspection and thus can be used for any customer data (web applications, logs, etc.). It is also generally set with permissions so that the customer has full access to the filesystem as well.
It can be, at extra cost. Please visit the Backup and Restore FAQ.