VPS performance: how much to hang in grams?

The hosting market has gradually moved from renting virtual sites to fully sane virtual servers, when there are guaranteed capacities and full access to the system and hardware. This strongly segmented the market: you can meet both tariff plans for 200-300 rubles, and more than 2000 rubles per month. In this case, all hosting providers offer more or less the same equipment performance (processor, memory, disk) and the reasonable question arises: what is the actual difference?

In order to sort out the issue with all the necessary depth, we conducted performance testing of more or less identical tariff plans from the most popular Russian providers on one of the most popular commercial platforms in Russia for creating websites - 1C-Bitrix. Why on it? There are a lot of reasons: both the standard installation of the basic site, and the availability of “specially selected” tariff plans, and simply the existence of a certain set of Beatrix Environments packages, which guarantee the relatively equal participation of various sites in testing (when we can compare performance without system or server environment).

Measurement Technique

For the technique, we used three basic performance parameters, partially related to each other:

• performance evaluation of 1C-Bitrix (everywhere used by Small Business Edition with the same site template and the same settings of the online store installation "out of the box"), the best of three consecutive measurements was taken;
• the average server response time across several characteristic pages (for example, main or product card, the same for all sites) without network costs (only waiting time for a response for a request without taking into account the request transmission time), the average time per day (or more );
• and the total load time of the same pages (it’s already essential here that the same site template was used for all sites), the average time per day (or more) was taken.

The choice of each parameter is clear: it reflects the actual performance of the site with one of the important components for the business (either an assessment of the site’s own server’s internal characteristics, or a real assessment of server performance on real requests — fault tolerance, or a real assessment of client performance — download speed)

All indicators were normalized to a “medium-good” value (for example, for the 1C-Bitrix estimate this was a reference 30), then they were multiplied and normalized to the price of the tariff plan. The latter is also quite obvious: we do not just find the most productive platform of all, but choose the most practical one for solving specific business problems (ensuring the resiliency and speed of the site).
')

Tariff Plans

The following tariff plans for the VPS were chosen for the study. The table also provides an estimate of the performance of 1C-Bitrix (measured without any load on the site).

Provider / rate	CPU (MHz)	Memory (MB)	Disk (GB)	Price, rub / month	Bitrix
Agava Optimal Bitrix	1000	512	12	1249	73.05
Rusonyx VPS Plus 2012	2000	1024	ten	499	66.13
Reg.Ru VPS-3	2000	1024	15	595	64,56
Selectel bitrix-m	1024	1024	50	750	58,82
HC ProVPS-3	2000	1024	ten	1400	58.37
Masterhost "Practical: 1C-Bitrix"	800	1024	ten	1700	55,69
FirstVDS VDS Acceleration	600	768	sixteen	249	38.05
FastVPS OVZ-5	2400	2000	14	671	29.01
TimeWeb VPS-Ultra	1500	1024	15	2500	14.63
Infobox VPS-1024 Linux	1000	1024	60	1100	11.48
IHC Earth	2000	1024	50	850	5.55

results

The WEBO Software score, as noted above, was formed as an integral value of the ratio between four factors: 1C-Bitrix performance evaluation, server response time (for different pages, the same between the sites under test), full load time for these same pages and by price.

Provider / rate	Bitrix	Response time (ms)			Download time (ms)			Price, rub / month	WEBO Software Evaluation
		one	2	3	one	2	3
Rusonyx VPS Plus 2012	66.13	82	78	60	1089	1208	855	499	536.87
Reg.Ru VPS-3	64,56	70	67	53	1750	1275	1121	595	390.88
Agava Optimal 1C-Bitrix	73.05	55	57	37	1602	1289	1048	1249	288.01
FirstVDS VDS Acceleration	38.05	147	144	93	1600	1662	1840	249	220.91
Selectel bitrix-m	58,82	144	154	116	1594	1431	1395	750	117.81
HC ProVPS-3	58.37	73	97	72	2402	1852	1511	1400	85.22
FastVPS OVZ-5	29.01	203	150	170	1916	2285	1423	671	41.86
Masterhost "Practical: 1C-Bitrix"	55,69	158	103	148	2410	2120	1837	1700	35.62
Infobox VPS-1024 Linux	11.48	570	481	388	1876	1978	1889	1100	3.5
TimeWeb VPS-Ultra	14.63	278	245	205	2818	2717	1815	2500	3.12
IHC Earth	5.55	736	637	402	5040	3317	2858	850	one

findings

The leaders in terms of performance / price (more performance for the same money) stood out well from the total number of hosting providers: these are Rusonyx, Reg.Ru, Agave and FirstVDS. Moreover, if Agava got to the leaders due to a very powerful virtual “hardware” (it’s not very clear why: the stated characteristics are inferior to most of the contestants), then FirstVDS broke out only because of the low price for fairly good performance.

At the same time, both Reg.Ru and Rusonyx are strongly detached from other providers, providing very good performance for a small price - around 500-600 rubles per month. However, Rusonyx beats Reg.Ru due to its good client performance (the virtual server is initially tuned for high performance and Apache, nginx, PHP and MySQL configurations are optimized for this), slightly yielding to the server response time.

In general, there is only one advice - to go down the list and consistently choose a hosting provider who will like the ratio of other factors.

And as it turned out, the performance evaluation of 1C-Bitrix is ​​quite a good tool for measuring the real performance of the site, but this assessment does not always give the result corresponding to real performance. Therefore, when choosing between several sites, in addition to this indicator, you also need to focus on actual performance (server response time), which can be quickly measured using a large number of well-known tools.