Andrei YurkevichPresident and CTO at Altoros Systems, [email protected]
Startup StoryCloud vs. Traditional Hosting
SOCIAL DATINGSOCIAL DATINGBEN BOB
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $WEEK
0 users 0 usersCost: $250/month Cost: $0.04/hour
0 1.0 1.0 01
AWS EASTAS
SMALL
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
500 users 500 500 users 500
Infrastructure spending: $250 Infrastructure spending: $70
250 1.1 1.1 701
ASAWS EAST
SMALL
СRASHED
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
500 users 500 500 users 500
Infrastructure spending: $250 Infrastructure spending: $70
250 1.1 1.1 701
AS
SMALL
AWS EAST
LARGEСRASHED
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
1 000 users 500 1 000 users 500
Infrastructure spending: $250 Infrastructure spending: $275 $205
500 1.2 1.2 3452
ASAWS EAST
LARGE
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
5 000 users 4 000 3 000 users 2 000
Infrastructure spending: $250 Infrastructure spending: $275
750 1.3 1.3 6203
ASAWS EAST
LARGEСRASHED СRASHED
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
AS
5 000 users 4 000 3 000 users 2 000
Infrastructure spending: $250 Infrastructure spending: $275
750 1.3 1.3 6203
LARGE
AWS EAST
EXTRALARGE
AS
СRASHED СRASHED
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
10 000 users 5 000 6 000 users 3 000
Infrastructure spending: $500 $250 Infrastructure spending: $550 $275
1 250 1.3 1.4 1 1704
AWS EAST
EXTRALARGE
AS
250
AS
250
VERSION
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
15 000 users 5 000 10 000 users 4 000
Infrastructure spending: $500 Infrastructure spending: $550
1 750 1.4 1.5 1 7205
AS
250
AS
250
VERSION
AWS EAST
EXTRALARGE
СRASHED
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
15 000 users 5 000 10 000 users 4 000
Infrastructure spending: $500 Infrastructure spending: $550
1 750 1.4 1.5 1 7205
AS
250
AS
250
VERSION
EXTRA LARGE
AWS EAST
PERMANENT
L L
L L
SCALEСRASHED
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
25 000 users 10 000 20 000 users 10 000
Infrastructure spending: $500 Infrastructure spending: $550 + $46
2 250 1.5 1.5 + scale 2 3666
24h/d L L .LB
2h/dx2
AS
250
AS
250
СRASHEDL L
VERSION
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
25 000 users 10 000 20 000 users 10 000
Infrastructure spending: $500 Infrastructure spending: $550 + $46
2 250 1.5 1.5 + scale 2 3666
24h/d L L .LB
2h/dx2
AS
250
AS
250
LB
AS
AS
ASСRASHED
L L
VERSION
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
35 000 users 10 000 60 000 users 40 000
Infrastructure spending: $1 000 $500 Infrastructure spending: $550 + $92
3 250 1.5 1.6 2 9587
AS
AS
LB
AS
VERSION
24h/d
L L
L L .LB
4h/dx2
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
AS
AS
LB
AS
60 000 users 25 000 120 000 users 60 000
Infrastructure spending: $1 000 Infrastructure spending: $550 + $275
4 250 1.6 1.7 3 7838
СRASHED
24h/d
L
L L .LB
6h/dx4
L L L
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
AS
AS
LB
AS
60 000 users 25 000 120 000 users 60 000
Infrastructure spending: $1 000 Infrastructure spending: $550 + $275
4 250 1.6 1.7 3 7838
x8
СRASHED
24h/d
L
L L .LB
6h/dx4
L L L
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
10 000 users 50 000 300 000 users 180 000
Infrastructure spending: $550 + $35 Infrastructure spending: $550 + $550
4 835 1.6 + cloud 1.8 4 8839
24h/d
L L
L L .LB
6h/dx8
L
L L .LB
3h/dx1
SOCIAL DATINGSOCIAL DATINGBEN BOB
$ VERSION VERSION $MONTH
2 000 users 8 000 500 000 users 200 000
Infrastructure spending: $550 Infrastructure spending: $550 + $5500
5 385 1.7 2.0 10 93310
24h/d
L L
L L .LB
12h/dx40
L
L L .LB
3h/dx1
BADREPUTATION
SOCIAL DATINGSOCIAL DATINGBEN BOB
SOCIAL DATINGSOCIAL DATINGBEN BOB
Recommendations:
1) Choose dedicated server when you know your load exactly and when there is no big variations in traffic
2) Keep in mind that system upgrade takes time
3) Remember that performance issues may negatively affect the existing user base
1) Choose cloud hosting when the load is unknown, especially if there is a chance that it will be very high
2) Take advantage of cloud elasticity
3) Monitor system load and upgrade to a more powerful instance before you run out of the resources
4) Test scalability of the application before your users do that
Andrei YurkevichPresident and CTO of Altoros [email protected]
Answers to questions
Top Related