National Day holiday, there were two little things, one is I spent quietly 35birthday, the second is to play the next grafana'and praised the first day after the holiday Engineer M: very professional.
1, why should I play grafanait?
A few months ago I submitted a demand to the background data engineer M, his selection and deployment of grafanathis third party open source tools, just two or three days to complete this demand. This efficiency is very fast, so I M and grafanaimpressive. But I'm still not satisfied, M finished all the data submitted by both graphical and I want the data table style. In order to further integrate other data for statistical analysis, I need to manually export data, and write complex pythonscripts pandasto do statistical analysis, run at least once a week. I often emerge new data requirements, they are not big, but to go again needs assessment, process development and acceptance of the schedule, it seems quite troublesome.
These are still only count bedding, the fuse is real before the National Day holiday sprintwrap-up meeting mention a few of our data products will be integrated into grafanaa unified implementation. CTO told me if I mentioned know some sql statement, you can write your own. --what? I understand the point of acridine sql statement, this is not just a finishing notes "[summary] how to interact and common python sql statement and mysql" thing. I was in "sql much better than this, put your horse over ah," the dauntless state-_-||
National Day holiday coming up, it would play with children look at the pictures.
2, how a short time to get started grafana?
M Prior to the development of my needs, it has completed the grafanadeployment and set up 数据源. This time he created a dedicated practice with dashboardme and open editing rights. Special note, only the right to query the data source to edit my permission, no additions and deletions to change the permissions of this data source is safe. In addition, M existing implementation also allows for the first time to get started, I can copy or imitate. These are completely new to me than grafana prepared to take advantage of the zero place.
I do not want M has developed needs to re-implement it again, I want to achieve their data needs. - what data you want, to what style rendering, my mind needs yours for the asking.
Specific learning process is divided into three parts:
- A: knowledge and understanding of
grafanahow to add in the dashboard, edit charts and other modules, is to understand how to use this tool. - B: knowledge and understanding of the data source (database of our products), which table, which fields like there.
- C: How to achieve data sql statement needs to write.
Part A, found the two grafana 如何使用like the article, probably browse buttons. Others always shining finishing step drawing step by step, learning experience is not good. Most of the time, I have been directly tinker. This visualization tools open source, you can try it yourself quickly familiar.
Section B, when a party hand, allows engineers to help to describe all forms of export is also OK; M to me is a form of class definition file. By exploring Part A, I soon discovered, create a module on the dashboard, if you select the type of line chart, sql statement editing area for humans to choose a table name, it is a data table which will be clear; if you select a table type, and use select * from table_name limit 50, you can present the data part of the table, so the table which will clear the field.
Section C, M I realized before sqlthe statement alone to copy jupyter notebookin their own dismantling is more basic knowledge, then a little familiar with. A little trick is very important for readability sql statement novice, can directly reduce complexity. So I use the markdownfollowing syntax, grammar presented is very clear:
Three or more portions without sequentially. They are more interested in which module, which began on the first; the process can also be cross rotation. Then there is the demand by implementing their own data, repeated again and again to consolidate and gain three or more parts, until my own satisfaction generated output. Put in the time overhead 24~48Henough friends, do not delay the holidays with the family, sleep late, watch movies.
3, to keep a clear division of labor in the workplace
Finally, it should emphasize that, I know he does not want to replace engineers to complete the development of the background data. For example:
1, a number of complex needs, I wrote it myself sql, it is difficult, it is also slow to learn. I will continue to collect those listed here, go schedule, please help engineers, not hard to dig into their own.
2, even if eventually I completed a lot of charts, but actually I consider only achieve, regardless of performance (also No ability to consider), so even if I write function, engineers also need to check and optimization.
Even so, I am in direct contact with the data source and hands-on with the grafanaimplementation, there are very significant benefits:
1, I know better raw data has been collected which, which indicators that I can define and statistics, which is the need to engineer further support.
2, some relatively simple, helpful to monitor business data / statistics, I can directly implement. No need to organize describe the demand - and engineers to communicate - to achieve the engineer to understand - I'll acceptance of such a complex process.
3, as the initiator of the demand, I demand that immature, yourself will be very efficient iterative process up.
Of course, these ideas are needed and engineers, communicate clearly superior, so as not to misunderstand acridine. If my notes to help you, then thumbs up or leave a message to tell me!