Hi, Our Appsheet Apps are running very slow...

(Srikanth S) #1

Hi,

Our Appsheet Apps are running very slow in India especially on mobile networks, probably because your Appsheet servers are all in US?

Relatively we see Youtube / Swiggy / Uber etc., load in any mobile phone in no time.

Is there anything that you can help us with?

(Dinh Nguyen Nguyen) #2

@Srikanth_S Please contact sales@appsheet.com for more information on the cost of increasing your thread count.

(Srikanth S) #3

Minimum time is like 25 seconds for loading our appsheet apps. Averages around 40-50 secs. Sometimes, it is as huge as 1.5-2 minutes.

(Srikanth S) #4

Account Id: 428375

(Dinh Nguyen Nguyen) #5

@Srikanth_S Hi, I’ve taken a brief look at the performance history of your some of apps. In most cases, the app takes a relatively long time to sync because there are quite a few tables in your app. With your current plan, your app will be allocated 2 computation threads, which means that your app can read at most 2 tables in parallel. This means if a table takes on average 3 seconds to sync, and there are 20 tables in the app, then the tables can be read in pairs of 2 that take 3 seconds each, resulting in a total sync time of 30 seconds.

Most of the apps in your account that I’ve seen typically have between 10 to 20 tables, and each table typically takes 2 to 4 seconds to load. As a result, the sync time that you described is more or less within the expected range.

Therefore, the cause of the problem is not because AppSheet servers are located in the US, but because your apps have a relatively large number of tables. If you want to improve the apps’ performance, you can either upgrade your plan to have more computation threads, or you can break your apps down into smaller apps that have fewer tables. The second approach is highly recommended.

Also, I do not recommend moving your data to a SQL database to solve this problem. As I have said, the problem is the large number of tables, not the amount of data in each table. Most of the tables in your apps don’t have a lot of data. However, since there’s an overhead cost incurred every time a table is read, having a lot of small tables will still result in longer sync time. Moving the tables to a SQL database won’t eliminate the overhead and thus won’t fix the problem. SQL databases should only be used if you have very large tables (at least a few thousand rows). For large tables, the overhead cost will cease to be significant relative to the time taken to read the rows in the table.

(Praveen Seshadri (AppSheet)) #6

Surely some of the tables are read-only. If so, enabling server caching will also help a lot.

(Srikanth S) #7

Thanks both. I have enabled Server Caching in my apps, which helped me a lot on the sync time. Could you also please send me an email on the costs if I increase number of tables to be read in parallel to a higher number?