Seminar "Bitrix High School" - as it were, presentations, video

Remember our old friend - a web developer and part-time system administrator? :) He, like many of our other readers, went on May 22 to the seminar “Development of high-loaded web projects: how to withstand millions of hits per day, so that everything works and nothing“ falls ” , which we announced even on Habré.

After that, it seems to us, it became a little easier for him to live, and the development and maintenance of large “heavy” projects became more understandable.

* * *
')
Almost half of the registrations for the seminar (and, apparently, for the participants themselves) were from Habra. We are very pleased that the topic was in demand.

Many people personally and on Twitter asked if presentations and videos would be published ...

Yes of course!

We are pleased to share all the materials with you!

1. "Development Platform for High-Load Web Service: Debugging Tools and Scalability" (Alexander Demidov, 1C-Bitrix)

Vertical scaling capabilities (by increasing memory, CPU, disks of an existing server) of a web project are always limited by physical characteristics. And sooner or later, the developer is faced with the task of implementing horizontal scaling - distributing the project to several servers. A traditional website is bad for this purpose - most often, scaling means that the project code is refined. In the report, we will describe how to provide scalability directly in the development platform, thereby making life easier for web developers. In addition, we will consider which tools for debugging, monitoring, and searching for potential bottlenecks can be used directly within the site, as they can complement external services.

• Presentation (pptx, 2.4 Mb)
• Video (flv, 419 MB)

2. “Construction of the system landscape of a high-load project” (Vitaly Gavrilov, Lenvendo)

An action plan and professional advice on building a system landscape that ensures the availability of high-load Internet projects:

• Hardware, what and what to choose.
• Options for building scalable solutions.
• Receiving load balancing and database resiliency.
• Examples of network traffic distribution.
• Backup.

• Presentation (ppt, 3.8 Mb)
• Video (flv, 183 MB)

3. "The use of its own methods of load testing in the development of high-load web projects" (Mikhail Tokovinin, QSOFT)

• Method of load testing used in QSOFT
• Setup and optimization of the project at the design stage
• Evaluation of load limits at the design stage

• Presentation (ppt, 870 Kb)
• Video (flv, 330 MB)

4. “Operation, performance monitoring, backups” (Ilya Pyatin, LineMedia)

• How to raise the server before it fell?
• How to start a project if on the first day you receive 100,000 users?
• Is it really necessary to always make backup?
• Why do I need a cluster if the load is small?
• Cloud cluster, pros and cons.

• Presentation (pptx, 1.2 Mb)
• Video (flv, 414 MB)

5. “Proactive monitoring and trend analysis” (Alexander Serbul, “1C-Bitrix”)

A web cluster with multiple application servers, a balancer, and a fault-tolerant configuration of relational storage is complex. In addition to properly setting up an application to achieve success, it is necessary to create and maintain an adequate system of monitoring and forecasting, to understand numbers, trends, because, as you know, the devil is hiding in trifles. What nagios tests are needed in the first place, which munin graphics you should pay attention to, why and how to analyze logs, how to write an emergency handler (handler nagios) and keep the web cluster in balance - let's consider these and many other issues.

• Presentation (pptx, 3.8 Mb)
• Video (flv, 387 MB)

6. “Exposing cloud magic” (Nikolay Dvas, Clodo.ru)

Marketing, leading the technology ahead, spawned a set of expectations from the cloud infrastructure. These expectations include:

• payment for the consumption of resources, no subscription fees and long-term contracts;
• unlimited scalability;
• the ability to obtain any number of resources at the moment when they are required, upon request, automatically;
• fault tolerance due to abstraction from iron.

The reality is significantly different from marketing expectations. It turns out that in the cloud for the client are very different aspects. In particular:

• savings are much more important than “fair” consumption pricing;
• scaling often does not work as expected and is not achieved by simply changing the hosting;
• fault tolerance is also not achieved by itself: there is a layer, which the hoster is not able to provide;
• “Resources by request” is a half-truth: not all the necessary resources can be requested and received;

The report focuses on adjusting popular myths about clouds, describing how to achieve cost-effectiveness, performance and fault tolerance, and listing what is extremely difficult to achieve.

• Presentation (pptx, 1.8 Mb)
• Video (flv, 242 MB)

7. “The architecture of the project“ Bitrix24 ”: how to make everything fly and not fall” (Alexander Demidov, “1C-Bitrix”)

One of the most important tasks that must be solved by any “cloud” provider is to gain the trust of service users. If we are talking about a business application, for example, CRM, intranet, mail, then this service should always be available - 7 days a week, 24 hours a day. Unstable work, failures, unloaded pages, a project closed for the period of planned service work are unacceptable. In addition, all customer data should be as secure as possible. The SaaS service provider should provide all of its users with the most secure environment possible. Using the example of a live Bitrix 24 service, the report describes the architecture of the project, principles of its automatic scaling depending on the load, integration mechanisms with cloud storages, specifics of using Amazon services, nuances of using MySQL database in a geographically distributed web cluster.

• Presentation (pptx, 4.1 Mb)
• Video (flv, 261 MB)

* * *

In the comments to the questionnaires that we collected after the seminar, when answering the question what was most useful, they often wrote: “atmosphere” and “experience”.

We had a very good and useful event. :) Thanks to everyone who participated!