We continue the article, the purpose of which is to share experience and show key features and frequent mistakes arising in the design and organization of power supply subsystems of IT infrastructure and data center in general. But I would like to slightly expand the audience and devote several sections to basic elements of electrical safety and protection of equipment and people.
 
 
Those who missed the first part or wants to remember the first part of you can go here .
 
 
For those who understand what an automaton and RCD are, for which they are needed, what they are protecting from - ...

<{full}>

 
We began to update the monitoring service for PgBouncer in our service and decided to do a little bit of combing. To make everything work, we draw the most well-known monitoring performance methodologies: Brendan Gregg's USE (Utilization, Saturation, Errors) and Tom Wilkie's RED (Requests, Errors, Durations).
 
Next, you will learn how we are all up there and about the features of the PgBouncer configuration parameters.
 
First about the methods themselves
 
Although these methods are quite well-known (about <{full}>

 
We ...

The author of the e-book is Eddie Osmani, one of the leaders of the development of Google Chrome
 
 

tl; dr

 

Image compression should always be automated

 
Optimization of graphics must be automated. It's easy to forget about it, recommendations change, and the content itself can easily slip past the assembly line. For automation when assembling, use imagemin or libvips . There are many others.
 
 
Most CDNs (for example, Akamai ) And third-party solutions like ...[/h][/h]

Until now, the problem of accumulation (accumulation) of electricity existed without any global changes, a technological breakthrough in this area no one could offer. It was the same battery Gaston Plante 1859 , modified by various auxiliary technologies and equipped with improvements in the field of electrochemical processes, the fight against the release of hydrogen, other materials of the hull, etc. At the turn of the beginning of 2000 appeared a different kind of ...

technique.   A few months ago, in the course of working on another project, the guys from our research laboratory conducted research on NVMe disks and software solutions in order to find the best option for building a software array.  The results of the tests were surprisingly discouraging at that time - the huge potential of the speed of NVMe disks did not correlate with the demonstrated performance of the available software products.  report G2M   At the moment, Intel is one of the leaders in the production of NVMe components. It was on their equipment that we conducted tests to evaluate the capabilities ...

Solutions for working with RAID arrays from hard drives have been used for a long time. In general, they continue to be popular in many areas, when a relatively inexpensive, fault-tolerant, large-capacity array is required. Given the size of modern hard drives, their speed, as well as other reasons, the greatest practical interest is arrays of RAID6 (or RAID6? if there are many disks). But this type of arrays has a low performance on random writes and it's not easy to do anything with it.
 
 
Of course, in this case we are talking about the speed of the "raw volume". In real life, it is added to ...

The proc file system (hereinafter simply procfs) is a virtual file system that provides information about processes. It is a "perfect" example of the interfaces of the following "everything is a file" paradigm. Procfs was developed a long time ago: at a time when servers on average served several dozen processes, when opening a file and subtracting information about the process was not a problem. However, time does not stand still, and now servers serve hundreds of thousands, or even more processes at the same time. In this context, the idea of ​​"opening a file for each process to subtract the ...

The material, the translation of which we publish today, is devoted to a story about how Airbnb optimizes the server parts of web applications with an eye to the increasing use of server rendering technologies. For several years, the company has gradually transferred its entire frontend to uniform architecture, according to which web pages are hierarchical structures of React-components, filled with data from their API. In particular, in the course of this process there was a planned abandonment of Ruby on Rails. In fact, Airbnb plans to switch to a new service based solely on Node.js, thanks ...

Hello. As an introduction, I want to tell you how I came to such a life.
 
 
Before I met Big Data and Spark, in particular, I had many and often optimized SQL queries, first for MSSQL, then for Oracle, and now I'm faced with SparkSQL.
 
 
And if there are already a lot of good books describing the methodology and "handles" for DBMS that can be twisted to get the optimal query plan, I have not seen such books for Spark. I came across more articles and sets of practices, more related to working through the RDD /Dataset API, rather than pure SQL. For me, one of the reference books on SQL optimization ...

Some time ago, one of my colleagues said that the place in DC is running out, there is nowhere to place the server, and the load is growing and it's unclear what to do, and probably will have to change all available servers to more powerful ones.
 
 
At that time I was busy with the task of drawing up optimal schedules, and I thought - what if using optimization algorithms to increase server utilization in DC? Hence the project was born, which I want to write about.
 
 
For the advanced, I will immediately say that in this article we will talk about bin packing, and the rest, who wants to learn ...