Monitoring in the data center: how we changed the old BMS to the new one. Part 1

What is BMS

The system for monitoring the operation of engineering systems in a data center is a key element of the infrastructure that directly affects such an important indicator for a data center as the speed of personnel response to emergencies and, consequently, the duration of uninterrupted operation. 

BMS (Building Monitoring System) monitoring systems are offered by many global vendors of data center equipment. During the work of Linxdatacenter in Russia, we had a chance to get acquainted with different systems and face diametrically opposed approaches of vendors to the operation of these systems. 

Here's how we completely upgraded our BMS over the past year and why.  

The root of the problem

It all started 10 years ago with the launch of the Linxdatacenter data center in St. Petersburg. The BMS system, according to the industry standards of those years, was a physical server with installed software, accessed through a client program (the so-called "thick" client). 

There were few companies offering such solutions on the market at that time. Their products were the standard, the only answer to an existing request. And we must give them their due: both then and today, market leaders as a whole cope with their basic task - the supply of functional solutions for the operation of data centers. 

The logical choice for us was a BMS solution from one of the world's largest manufacturers. The selected system at that time met all the requirements for monitoring a complex engineering facility, which is a data center. 

However, over time, the requirements and expectations of users (that is, us data center operators) from IT solutions have changed. And large vendors, as the market analysis of the proposed solutions showed, were not ready for this.

The corporate IT market has experienced a serious influence of the B2C sphere. Digital solutions today should provide the convenience of the end user - this is the goal set by the developers. This is evident in the improvement in user interfaces (UI) and the quality of user experience (UX) of many enterprise applications. 

A person gets used to comfort in everything related to digital tools in everyday life, and makes the same demands on the tools that he uses for work tasks. People expect from enterprise applications the same visibility, intuitiveness, simplicity and transparency that they get in financial services services, calling a taxi or online shopping. IT professionals implementing solutions in a corporate environment also strive to get all the modern "goodies": simple deployment and scaling, fault tolerance and unlimited customization options. 

Major international vendors often overlook these trends. Based on their many years of authority in the industry, corporations in dealing with customers often turn out to be categorical and inflexible. The illusion of their own indispensability does not allow them to see how young technology companies appear literally under their noses, offering alternative solutions tailored to a specific customer, and without overpaying for the brand.

Disadvantages of the old BMS system 

The main disadvantage of the existing outdated BMS solution for us was its slow operation. Investigating several events related to the insufficiently fast response of the staff on duty helped us understand that sometimes events were displayed in the BMS system with a long delay. At the same time, the system was not overloaded or faulty, just the versions of its components (for example, JAVA) were outdated and could not work correctly with new versions of operating systems without updates. They could only be updated together with the BMS system, while the vendor did not provide automatic version succession, that is, for us, the process would be almost as laborious as switching to a new system, and the new solution retained some of the shortcomings of the old one.  

Let's add a few more unpleasant "little things" here:

1. Fee for connecting new devices on the principle of "one IP address - one paid license"; 
2. Inability to update the software without purchasing a support package (meaning updating free components and fixing errors in the BMS program itself);
3. High cost of support; 
4. Location on a "hardware" server that can fail and has limited computing resources;
5. "Reservation" by installing a second iron server, with a duplicate license package. At the same time, there is no synchronization of databases between the main and backup servers, which means manual transfer of the database and a long time for the transition to the backup;
6. "Thick" user client, inaccessible from the outside, without extension for a mobile device and remote access options;
7. A truncated web interface without graphic cards and sound notifications, accessible from the outside, but practically not used by employees due to its lack of information;
8. Lack of animation in the interface - all graphics consist only of a background image and static icons. As a result - a general low level of visibility;

   Everything looked like this:

   
   
   

9. The limitation in creating virtual sensors is that only the addition function is available, while the models of real sensors require the ability to perform a set of mathematical operations for correct calculations that reflect the realities of work; 
10. Impossibility to obtain data in real time or from the archive for any purpose (for example, to display in the client's personal account);
11. The complete lack of flexibility and the ability to change something in BMS for existing data center processes. 

Requirements for a new BMS system

Considering the above, our main requirements are as follows:

1. Two independent mutually reserving machines with automatic synchronization, running on two different cloud platforms in different data centers (in our case, Linxdatacenter data centers St. Petersburg and Moscow);
2. Free addition of new devices;
3. Free updating of software and its components (with the exception of functional improvements);
4. Open source code, which allows us to independently support the system in case of problems on the developer's side;
5. The ability to receive and use data from BMS, for example, on a website or in a personal account;
6. Access through a WEB browser without a "thick" client;
7. Using domain accounts of employees to access BMS;
8. The presence of animation and many more small and not very wishes that materialized in a detailed TOR.

Last straw

At the moment when we realized that the data center had outgrown its BMS, the most obvious solution seemed to us to upgrade the existing system. “They don’t change horses in midstream,” right? 

However, large corporations, as a rule, do not offer custom modifications to their decades-long polished solutions sold in dozens of countries. While young companies are testing an idea or a prototype of a future product on potential consumers and relying on user feedback in product development, corporations continue to sell licenses for a once really cool, but, alas, today outdated and inflexible product.

And we felt the difference in the approach for ourselves. In the course of correspondence with the manufacturer of the old BMS, it quickly became clear that the upgrade of the existing system proposed by the vendor would actually result in the purchase of a new system for us with a semi-automatic transfer of the base, high cost and pitfalls during the transfer that even the manufacturer himself could not predict. Of course, in this case, the cost of technical support for the updated solution grew, and there was a need to buy licenses when expanding.

Worst of all, the new system could not fully meet our redundancy requirements. The updated BMS system could be implemented, as we wanted, on a cloud platform, which would allow us to abandon the hardware, but the reservation option was not included in the price. To back up the data, we would have to buy a second virtual BMS server and an additional set of licenses. With a cost of one license of about $76 and a number of IP addresses of 1000 units, there is an additional $76 in additional costs for licenses for a backup machine alone. 

The highlight of the new version of BMS is the need to purchase additional licenses “for all devices” – even for the main server. Here it is necessary to clarify that there are devices connected to the BMS through gateways. The gateway has one IP address but controls multiple devices (average 10). In the old BMS, this required one license per gateway IP address, the statistics looked something like this: "IP addresses / licenses 1000, devices 1200". The updated BMS worked on a different principle and the statistics would look like this: “IP addresses 1000, devices / licenses 1200”. That is, the vendor changed the principle of assigning licenses in the new version, and we had to buy an additional 200 more licenses. 

The budget for the "upgrade" eventually consisted of four items: 

• the cost of the cloud version and services for migration to it; 
• additional licenses to the existing package for devices connected via gateways;
• the cost of the backup cloud version;  
• a set of licenses for a backup machine. 

The total cost of the project was over $100! And that's not to mention the need to purchase licenses for new devices in the future.

As a result, we realized that it would be easier for us - and maybe even cheaper - to order a system created from scratch, taking into account all our requirements and providing for the possibility of upgrading in the future. But those who wanted to develop such a complex system still needed to be found, compare proposals, choose and go with the finalist the path from technical specifications to implementation ... Read more about this in the second part of the material very soon. 

Source: habr.com