Every WebSocket result, side by side.

7 Linode servers, from 1 to 8 vCPUs, shared and dedicated CPU, API Maker with its MongoDB and Redis on the same machine, and 3 clusters of API servers which share one Redis. 50,167,783 clients let in, 214,875,630 notifications delivered, 28,019,607 HTTP calls, 0 failures. Each number links to its run, with the output.

Most clients on one server 3,662,564 8 shared vCPUs, 32 GB

Memory per client 8.5 to 8.8KB 118,483 to 123,507 clients per GB

New clients / second 2,602 to 10,939 both tokens and a subscription

Delivery p50 0.83 to 2.51ms a notification for one person

Clients connected at once¶

WebSocket clients connected at onceShared CPUDedicated CPU

1 vCPU · 1 GBshared · $5 / month26,484

2 vCPU · 4 GBshared · $24 / month338,016

4 vCPU · 8 GBshared · $48 / month801,972

8 vCPU · 32 GBshared · $192 / month3,662,564

2 vCPU · 4 GBdedicated · $36 / month338,006

4 vCPU · 8 GBdedicated · $72 / month795,888

8 vCPU · 16 GBdedicated · $144 / month1,732,035

Server	CPU · price / month	Clients connected	What ended first	API Maker	Whole server, used	Per client	Clients per $ a month
1 vCPU · 1 GB	Shared · $5	26,484	memory	126 MB	813 MB of 961 MB	8.6 KB	5,297
2 vCPU · 4 GB	Shared · $24	338,016	memory	1,110 MB	3,417 MB of 3,916 MB	8.5 KB	14,084
4 vCPU · 8 GB	Shared · $48	801,972	memory	2,495 MB	6,937 MB of 7,941 MB	8.7 KB	16,708
8 vCPU · 32 GB	Shared · $192	3,662,564	memory	10,196 MB	27,618 MB of 32,096 MB	8.8 KB	19,076
2 vCPU · 4 GB	Dedicated · $36	338,006	memory	1,108 MB	3,465 MB of 3,916 MB	8.5 KB	9,389
4 vCPU · 8 GB	Dedicated · $72	795,888	memory	2,474 MB	6,917 MB of 7,941 MB	8.7 KB	11,054
8 vCPU · 16 GB	Dedicated · $144	1,732,035	memory	5,179 MB	13,942 MB of 15,992 MB	8.8 KB	12,028

Memory¶

With every client connected	1 vCPU shared · 1 GB	2 vCPU shared · 4 GB	4 vCPU shared · 8 GB	8 vCPU shared · 32 GB	2 vCPU dedicated · 4 GB	4 vCPU dedicated · 8 GB	8 vCPU dedicated · 16 GB
Clients	26,484	338,016	801,972	3,662,564	338,006	795,888	1,732,035
API Maker (process)	126 MB	1,110 MB	2,495 MB	10,196 MB	1,108 MB	2,474 MB	5,179 MB
Redis	30 MB	356 MB	808 MB	3,533 MB	356 MB	802 MB	1,633 MB
Linux kernel (slab)	200 MB	1,635 MB	3,715 MB	16,443 MB	1,627 MB	3,691 MB	7,839 MB
MongoDB of API Maker	110 MB	117 MB	64 MB	61 MB	125 MB	66 MB	52 MB
Sandboxes	143 MB	91 MB	206 MB	245 MB	99 MB	203 MB	334 MB
Whole server, used	813 MB	3,417 MB	6,937 MB	27,618 MB	3,465 MB	6,917 MB	13,942 MB
Available	149 MB	499 MB	1,005 MB	4,478 MB	451 MB	1,024 MB	2,050 MB
Swap used	176 MB	310 MB	609 MB	2,691 MB	270 MB	594 MB	1,525 MB
API Maker at rest, no client	103 MB	110 MB	118 MB	121 MB	109 MB	118 MB	117 MB
Per client : API Maker	3.3 KB	3.1 KB	3.3 KB	3.2 KB	3.1 KB	3.3 KB	3.2 KB
Per client : Linux kernel	4.3 KB	4.3 KB	4.4 KB	4.5 KB	4.3 KB	4.4 KB	4.5 KB
Per client : Redis	1.0 KB	1.0 KB	1.1 KB	1.1 KB	1.0 KB	1.1 KB	1.1 KB
Per client : in all	8.6 KB	8.5 KB	8.7 KB	8.8 KB	8.5 KB	8.7 KB	8.8 KB
Clients per GB	121,927	123,507	120,250	118,483	123,507	120,250	118,886

New clients per second¶

100 clients at a time	1 vCPU shared · 1 GB	2 vCPU shared · 4 GB	4 vCPU shared · 8 GB	8 vCPU shared · 32 GB	2 vCPU dedicated · 4 GB	4 vCPU dedicated · 8 GB	8 vCPU dedicated · 16 GB
Connect : the token of the API user	5,905	9,799	15,228	24,511	7,417	13,849	21,780
Connect : both tokens	5,302	9,504	12,826	24,719	7,077	12,535	19,391
Connect with both tokens and subscribe	2,602	4,041	7,009	10,939	3,464	6,627	10,093
Subscribe to an event with a can user connect code	2,097	3,438	5,916	9,280	2,891	5,533	8,963
Subscribe : the code reads the database	959	1,605	2,735	4,381	1,368	2,046	3,919
Subscribe : the code reads the clock	1,960	3,031	5,243	7,741	2,550	4,505	7,543
Subscribed after, p50	33.2 ms	21.5 ms	12.5 ms	7.91 ms	25.1 ms	13.4 ms	8.85 ms
Subscribed after, p99	47.5 ms	31.2 ms	18.9 ms	13.4 ms	35.0 ms	19.3 ms	12.8 ms

Notifications¶

How late : at a pace	1 vCPU shared · 1 GB	2 vCPU shared · 4 GB	4 vCPU shared · 8 GB	8 vCPU shared · 32 GB	2 vCPU dedicated · 4 GB	4 vCPU dedicated · 8 GB	8 vCPU dedicated · 16 GB
A notification for one person : delivery p50	2.51 ms	1.45 ms	1.06 ms	0.94 ms	1.74 ms	1.04 ms	0.83 ms
… p99	3.82 ms	3.15 ms	2.03 ms	6.54 ms	3.70 ms	2.03 ms	1.27 ms
A broadcast : delivery p50	10.3 ms	8.82 ms	7.56 ms	8.70 ms	10.3 ms	8.00 ms	6.53 ms
… p99	21.6 ms	18.0 ms	16.4 ms	19.9 ms	18.3 ms	17.3 ms	14.0 ms
Rows inserted in a table : delivery p50	51.3 ms	25.3 ms	18.1 ms	13.4 ms	35.1 ms	20.7 ms	12.9 ms
… p99	103 ms	44.4 ms	30.0 ms	35.5 ms	54.8 ms	34.5 ms	19.8 ms

How many : notifications a second	1 vCPU shared · 1 GB	2 vCPU shared · 4 GB	4 vCPU shared · 8 GB	8 vCPU shared · 32 GB	2 vCPU dedicated · 4 GB	4 vCPU dedicated · 8 GB	8 vCPU dedicated · 16 GB
A notification for one person	5,216	7,562	13,441	21,208	6,618	12,791	21,001
… HTTP calls a second	5,285	7,657	13,711	21,628	6,757	13,049	21,252
A broadcast	72,245	103,387	147,974	196,995	69,874	117,698	206,570
… HTTP calls a second	73	105	150	199	71	120	209
Rows inserted in a table	40,498	59,275	92,279	136,942	49,312	76,505	132,499
… HTTP calls a second	416	604	932	1,396	506	779	1,350

With every client connected¶

	1 vCPU shared · 1 GB	2 vCPU shared · 4 GB	4 vCPU shared · 8 GB	8 vCPU shared · 32 GB	2 vCPU dedicated · 4 GB	4 vCPU dedicated · 8 GB	8 vCPU dedicated · 16 GB
Clients connected	26,484	338,016	801,972	3,662,564	338,006	795,888	1,732,035
A notification for one person : delivery p50	2.49 ms	1.47 ms	1.22 ms	1.28 ms	1.76 ms	1.03 ms	0.87 ms
… p99	7.48 ms	3.27 ms	2.44 ms	121 ms	3.03 ms	1.99 ms	1.47 ms
Notifications a second, one person each	5,226	6,967	10,799	8,384	5,705	9,745	15,938
Get all with both tokens, no client : req/s	2,095	3,379	6,072	9,529	3,058	5,268	9,672
Get all with both tokens, every client connected : req/s	1,983	3,006	4,767	3,997	2,614	4,185	7,305

A restart¶

API Maker restarts	1 vCPU shared · 1 GB	2 vCPU shared · 4 GB	4 vCPU shared · 8 GB	8 vCPU shared · 32 GB	2 vCPU dedicated · 4 GB	4 vCPU dedicated · 8 GB	8 vCPU dedicated · 16 GB
Clients connected	19,863	253,512	601,479	–	253,504	596,916	–
All back after	10.0 s	90.0 s	189 s	–	159 s	189 s	–
A client was away, p50	7.4 s	48.9 s	76.7 s	–	68.7 s	79.1 s	–
A client was away, p99	9.1 s	78.2 s	173 s	–	96.5 s	172 s	–
Every client leaves at once : sockets closed after	1.5 s	10.1 s	18.0 s	–	15.1 s	23.4 s	–
… Redis forgot them after	2.6 s	13.8 s	26.9 s	–	16.1 s	29.1 s	–

Several servers¶

The same 8 dedicated vCPUs and 16 GB for API Maker, as one server, two, and four, with MongoDB and Redis on a server of their own : Scale to N users.

	1 × 8 vCPU · 16 GB	2 × 4 vCPU · 8 GB	4 × 2 vCPU · 4 GB
Clients connected at once	2,022,026	1,915,960	1,619,372
… on each server	2,022,026	957,980	404,843
Memory per client, in all	8.8 KB	8.8 KB	8.6 KB
New clients a second (both tokens, a subscription)	12,412	12,696	10,297
… connect only, both tokens	19,917	21,417	17,882
A notification for one person : delivery p50	1.01 ms	1.18 ms	1.37 ms
… p99	1.72 ms	1.80 ms	2.51 ms
Notifications a second, one person each	24,305	24,977	25,566
Notifications a second, broadcast	217,711	228,910	272,609
One server stops : clients dropped	1,516,519 of 1,516,519 (100 %)	383,192 of 766,384 (50 %)	242,906 of 971,623 (25 %)
… all back after	221 s	68.0 s	38.0 s

The servers¶

	1 vCPU	2 vCPU	4 vCPU	8 vCPU	2 vCPU	4 vCPU	8 vCPU
Plan	Nanode 1 GB g6-nanode-1	Linode 4 GB g6-standard-2	Linode 8 GB g6-standard-4	Linode 32 GB g6-standard-8	Dedicated 4 GB g6-dedicated-2	Dedicated 8 GB g6-dedicated-4	Dedicated 16 GB g6-dedicated-8
CPU	AMD EPYC 7642 shared cpu	AMD EPYC 7642 shared cpu	AMD EPYC 7642 shared cpu	AMD EPYC 7642 shared cpu	AMD EPYC 7642 dedicated cpu	AMD EPYC 7642 dedicated cpu	AMD EPYC 7642 dedicated cpu
Memory	1 GB	4 GB	8 GB	32 GB	4 GB	8 GB	16 GB
Price / month	$5	$24	$48	$192	$36	$72	$144
Load generator	Dedicated 8 GB	Dedicated 8 GB	Dedicated 16 GB	Dedicated 64 GB	Dedicated 8 GB	Dedicated 16 GB	Dedicated 32 GB
… busy at most	66 %	98 %	91 %	85 %	98 %	90 %	84 %

How every number was taken : How the WebSocket benchmarks are made. The data : benchmark dataset (https://apimaker.dev/json/benchmark_dataset.json).