This was written by Jennifer Davis.
For most of us, the end of year brings performance appraisals and reflection on the year's setbacks and accomplishments both professional and personal. In the glass bowl of life, I ask myself if I've made a difference, earned respect, have I grown from where I started the year? Technology has brought us innovation to hack our lives and measure our personal success through metrics. For example measuring sleep, weight, and activity. For example, I can see how much I slept, how fast I biked to work this morning, how often I biked, and the steps forward (or back) towards my goals. There is a dearth in tools available to measure personal work growth and effectiveness.
Being an effective system administrator requires an ability to do several (seemingly obvious but often rather fraught) things: To break down projects into actions that we understand as a part, as a whole, and can manage in a discrete period of time; explaining this roadmap to other teams; and successfully keeping implementation on schedule while being flexible enough to handle any issues that arise. The job descriptions and responsibilities of system administrators can vary greatly in scope and the corresponding degrees of difficulty and creativity necessary to succeed. Since "system administrator" alone can sometimes function as a vague catch-all for such a diversity of tasks and functions we use a variety of sometimes unwieldy names to better specify our roles and focus. Regardless of title there is a great deal of commonality in how teams we work for/with view us and depend upon our knowledge and skills. In some cases it's a bit like being a member of a symphony in which the strings, the brass, and the wind sections cannot agree upon the tempo or even what piece to play.
At a team level, management has a vision of what the team should be doing and how it should be working. Often our work is considered a cost center, something that doesn't produce a direct profit and is generally first in line for cuts. Management becomes focused on the bottom line to the detriment of building a strong team and an encompassing vision. Teams are put in the unfortunate position of competing for finite resources. For better or worse, the team that "markets" itself best generally comes out ahead.
Rather than embracing the frustration that may come with a sub-optimal allocation of time or resources, give your manager visuals to obtain the resources you need by showing value and maintain focus on the team's vision. Understand goals for the team, and know how your efforts fit into that vision. By broadening your focus a little bit you may have more influence in determining short term goals. Is there information that your manager is not aware of that may change short term goals? Is he or she unaware of the real-world time it takes to perform a particular task, or how project A by team X will impact the ability of you and your team to complete project B? Communicate information succinctly and clearly with your interpretation of the data. Present not just the problems but potential solutions that can solve the problem. Back your solutions with the metrics you collect about your own work.
The first part in creating visuals is to establish an efficient workflow that you can measure. The metrics you collect over time will convince others of the validity of your timetable and proposed project maps. To obtain metrics, break big projects into tasks. Break large tasks into smaller tasks. A task should be no more than 4 hours of work. The reality is that for most of us each 8 hour work day allows for 4 hours of work. The other 4 hours are taken up by interrupt tasks, networking with individuals from other teams that your work depends on, and communicating your status. Stop, take a deep breath, embrace this reality, and then take the time to ensure that those 4 hours of work are quality time.
Qualify each task. This qualification is dependent on your environment. Sample qualifications include prep, installation, deployment, monitoring, business continuity, education, documentation and mentoring. Allocate the 4 hours of interrupt tasks as a chunk of time labeled as "interrupt". Depending on your work, your job may be 100% interrupt driven. If this is the kind of work you enjoy doing then measuring your success will take more time. I measure my interrupts by total time taken and scale of interrupts focusing my measurements on non-interrupts. Ideally, my interrupt time is spent mostly on networking and mentoring versus manual domain specific tasks that only I can do.
Make an informed estimate of how much time and effort each task will take. Excessively rosy estimations can lead to unnecessary stress. Be conservative. Everyone has peaks and valleys in terms of attention and performance that do not always match up. For me, mornings are key to establish the work I'm going to complete for the day with a couple of hours of concentrated work. Around lunch time is my interrupt time and I try to schedule meetings then. I re-evaluate my progress for the morning and finish up the day with uninterrupted time. Each task should be between 1-4 hours. Order your tasks based on team goals and priorities. Sync up with your team and manager regularly to ensure that you are working on the right projects and tasks. To help keep my team in sync, I introduced a daily standup with the goal of 2 minutes max per person. Then in the weekly meeting with my manager, I asked about team goals verifying that my prioritized list maps out to these goals.
Project and Large Task State
Determine the states of work your projects can have. Depending on the tools in your environment the terms available may not map directly to your organization of tasks. Your personal term association matters more than the actual terms. I use "Accepted", "WIP", "Done", and "Validated". "Accepted" means I know about it, and have accepted that I own the responsibility of ensuring that the work is done. I understand the specific goals and a finish state. "WIP" means I've started progress on this task and it has status. It can involve further clarification of goals, and some amount of progress. "Done" means that I believe that the task has been completed. "Validated" means that I've verified that the work has been completed either through buddy checks, management acceptance, or some automated test to verify the completed state. In general I try to find an external method to verify that the work is complete.
Tracking the actual effort it takes to complete a task and the time each project takes to pass through states that you have defined helps you measure the time that it takes you to complete specific tasks and projects as well as where the time is spent across the project between tasks. Measure your completion of tasks and projects with your personal performance metrics. Think about the questions that you'd like your manager to understand. These questions should be specific to you, your work, and what value you bring to your job. Examples include: How long does it take to complete a repetitive manual task? How long would it take if you were given enough time to automate the manual task? What would the effect of fixing a minor bug in the application have on the overall time it takes for your team to complete manual task? Are you getting better at completing tasks by doing more and finishing them faster? Turn these metrics into graphs. Images convey meaning quickly. You can use R, YUI Charts, Google Spreadsheets with charts to make some nifty graphs.
Find or write tools that measure the completion of work. The data you collect about your work provides information for your status reports, requests for out of band raises, as well as end of year summary. Think through the tools and processes in your environment. If you use bugzilla or RT for example find the mechanisms that you can use to pull data straight from the source automatically. If you manage the bugzilla database, you can pull this information directly with mysql queries. If you don't investigate whether there is an API provided that you can poll for this information. Interesting metrics include total tasks per day, average latency per task, average latency per type of work, and percentage of time spent doing specific types of tasks. Note that if you are spending less than 4 hours on interrupt tasks consistently, make sure that you are spending adequate time reaching out to your peers in other groups. Networking and understanding relevance is part of your job.
Communicate the schedule and completion of tasks to everyone who is affected by the completion and delays of your work. Consistent clear communication builds your perceived value to management, team, and customers. Visuals give your manager effective views of your value to use at performance review time to ensure your salary increases.
Learn to say no
When you organize your priorities and tasks towards improving your efficiency and your manager's goals you have to learn to say no to some of the interrupts. This is not the BOFH "no". This is an opportunity to increase your value. I often struggle to say no when confronted with new and interesting problems or when someone needs help with a problem that would be very easy for me to do.
When someone asks you for something, is it something that you must do, or can you teach them how to help themselves? Depending on the urgency invest some time training the individual how to help themselves. Ideally you go to their workspace and let them run through the process with your guidance while keeping your hands off the keyboard. It's in the nature of a sysadmin to jump in and help, so I hold my hands behind my back. This is also important during emergencies to ensure that other's can be comfortable executing on common problems. Tell them to take notes and document the procedure for your review later. If the process is complicated you may need to run through this exercise twice. By investing the time in training someone, you multiply the effectiveness of getting this task done in the future. The next time someone needs help you can point them to the documentation and if you are busy to the person you previously trained. I have found that people get frustrated at first through this process but over time that they appreciate that they can help themselves and have more confidence in helping others through the same process.
If this is something you must do, what is the impact of you stopping and completing the request? Will the short term task unblock a big team objective? How complicated is it and how much time will it take you to do it? When you think of these questions this will give you the gut instinct of whether you should go ahead and get the work done. Understand the work you are doing, and own your time. A good sys admin knows when she must fail to deliver on a project due to unplanned interrupts with a higher priority. Management and key stakeholders should be informed of compromised deadlines with brief details about why the task was preempted. I find clear to the point communication difficult so I solve this by creating a for more information link in the email that points to a wiki. The wiki I can edit for clarity and more details as needed.
Modify your behaviors
Over time you will see a trend in your behaviors. You can see what you like about your job and whether you are spending time on the right kind of work. If you do not like what you see in your status reports, change the way you work or find a new job. At the end of the year you won't be feeling regret over missed opportunities because you have a way to keep track of your value over time and ensure that you are making progress.
Whatever kind of ops or administrator that your title or potential job reflects, look at the opportunities you have to work more efficiently. The effort you put into sending information out will decrease the number of times you are pulled away from the task-in-progress to be solicited for information piecemeal. Open up, network with others and you'll have more time to get deep into the parts of your job that increase your satisfaction, effectiveness, and bring your brain happiness.