AWS - WordPress

[ntayya]

WordPress site ekak thiyenawa, dedicated server ekaka thiyenne, eka AWS walata move karanne kohomada, with HA, autoscaling, and need to have proper security,

 

[NEMISIS]

Have you consider moving into a WP specific platform, like wp engine or so.

What is your actual requirement, or what are you trying to solve (improve performance, reduce the cost, reduce downtime etc)
Did you analyse your numbers/statistic, at-least what is the maximum number of concurrent users/requests you need to handle, what is your current up time etc.

My initial idea was, if it is running on a dedicated server without any issue, and if you need HA, try adding another node (simply start a copy of the existing server and use a single DB for both servers/use a managed db instance), and configure a load balancer. May be you can use cloud-flare load balancer and you will get all the benefits/security of the cloud-flare.

If you are going to implement a fully fledged HA/Auto scaling setup, it won't be a simple task, and running cost would be high compared to the current single dedicated server.
You can try to containerise your application and deploy on some managed container service.

If it is a wp site, 80% of the time, you won't need such complex setup.

I can give you a better answer if you can provide more details. Do not try to complicate things if it is not necessary.

 

[ntayya]

Have you consider moving into a WP specific platform, like wp engine or so.

What is your actual requirement, or what are you trying to solve (improve performance, reduce the cost, reduce downtime etc)
Did you analyse your numbers/statistic, at-least what is the maximum number of concurrent users/requests you need to handle, what is your current up time etc.

My initial idea was, if it is running on a dedicated server without any issue, and if you need HA, try adding another node (simply start a copy of the existing server and use a single DB for both servers/use a managed db instance), and configure a load balancer. May be you can use cloud-flare load balancer and you will get all the benefits/security of the cloud-flare.

If you are going to implement a fully fledged HA/Auto scaling setup, it won't be a simple task, and running cost would be high compared to the current single dedicated server.
You can try to containerise your application and deploy on some managed container service.

If it is a wp site, 80% of the time, you won't need such complex setup.

I can give you a better answer if you can provide more details. Do not try to complicate things if it is not necessary.

Yes but our one of our developer and the client want to move to AWS , because of onc incident, a downtime, but I don't have much experience with AWS and and cost is not predictable, Traffic 2000 unique visitors per day

 

[anonymouseninja]

AWS Lightsail for WordPress

 

[B.Chat]

Up

 

[NEMISIS]

Here is my suggestion

1. You don't need auto scaling for 2000 users ( that is 1 user per min on average, but in real life there will be some peak time, but still you don't need auto scaling at this point).
2. What you need is HA and way to monitor your setup.
3. Configure a manage DB instance with Zone redundancy ( this is very important and this will be bit costly too, may be you can configure a replica but these are costly things to), I think you can configure all of them in one go in AWS, if you are using AWS RDS by configuring Multiple availability zones with failover.
4. Configure two VMs/Dedicated servers in two different availability zones, deploy your WP app in these VMs and connect both of them to the db in step 3. If users are uploading files to your wp site, then you will have to configure some shared file system across VMs.
5. Configure LB to balance your load between two VMs ( you can use LB available in AWS or Cloudflare etc)
6. Point your DNS to CF( or whatever the service you are going to use)
7. You need a way of monitoring your setup, CF, Two VMs, and DB. There are multiple options for this, but i suggest to use existing services like Site24x7. Reason is, if you are critical about your uptime, you have to be critical about uptime of your monitoring service.
8. Implement some backup strategy.

Possible problem : If users are logging into your site, you might want to take a look at how to handle session. I guess WP depends on php sessions, which may store some data locally, if user hits one VM in one request and other VM in the next request, user might get logged out or face issues. You might be able to fix this with shared session storage, may be Redis or Memcached will help.

If you need HA and auto scaling, you have to run two K8S clusters in two zones at minimum. That would be very costly compared your current single VM. Without zone redundancy, you can't say you have HA.

 

[ntayya]

Here is my suggestion

1. You don't need auto scaling for 2000 users ( that is 1 user per min on average, but in real life there will be some peak time, but still you don't need auto scaling at this point).
2. What you need is HA and way to monitor your setup.
3. Configure a manage DB instance with Zone redundancy ( this is very important and this will be bit costly too, may be you can configure a replica but these are costly things to), I think you can configure all of them in one go in AWS, if you are using AWS RDS by configuring Multiple availability zones with failover.
4. Configure two VMs/Dedicated servers in two different availability zones, deploy your WP app in these VMs and connect both of them to the db in step 3. If users are uploading files to your wp site, then you will have to configure some shared file system across VMs.
5. Configure LB to balance your load between two VMs ( you can use LB available in AWS or Cloudflare etc)
6. Point your DNS to CF( or whatever the service you are going to use)
7. You need a way of monitoring your setup, CF, Two VMs, and DB. There are multiple options for this, but i suggest to use existing services like Site24x7. Reason is, if you are critical about your uptime, you have to be critical about uptime of your monitoring service.
8. Implement some backup strategy.

Possible problem : If users are logging into your site, you might want to take a look at how to handle session. I guess WP depends on php sessions, which may store some data locally, if user hits one VM in one request and other VM in the next request, user might get logged out or face issues. You might be able to fix this with shared session storage, may be Redis or Memcached will help.

If you need HA and auto scaling, you have to run two K8S clusters in two zones at minimum. That would be very costly compared your current single VM. Without zone redundancy, you can't say you have HA.

Thanks for your help, I will go through above steps

 

[Mr.Thor]

Here is my suggestion

1. You don't need auto scaling for 2000 users ( that is 1 user per min on average, but in real life there will be some peak time, but still you don't need auto scaling at this point).
2. What you need is HA and way to monitor your setup.
3. Configure a manage DB instance with Zone redundancy ( this is very important and this will be bit costly too, may be you can configure a replica but these are costly things to), I think you can configure all of them in one go in AWS, if you are using AWS RDS by configuring Multiple availability zones with failover.
4. Configure two VMs/Dedicated servers in two different availability zones, deploy your WP app in these VMs and connect both of them to the db in step 3. If users are uploading files to your wp site, then you will have to configure some shared file system across VMs.
5. Configure LB to balance your load between two VMs ( you can use LB available in AWS or Cloudflare etc)
6. Point your DNS to CF( or whatever the service you are going to use)
7. You need a way of monitoring your setup, CF, Two VMs, and DB. There are multiple options for this, but i suggest to use existing services like Site24x7. Reason is, if you are critical about your uptime, you have to be critical about uptime of your monitoring service.
8. Implement some backup strategy.

Possible problem : If users are logging into your site, you might want to take a look at how to handle session. I guess WP depends on php sessions, which may store some data locally, if user hits one VM in one request and other VM in the next request, user might get logged out or face issues. You might be able to fix this with shared session storage, may be Redis or Memcached will help.

If you need HA and auto scaling, you have to run two K8S clusters in two zones at minimum. That would be very costly compared your current single VM. Without zone redundancy, you can't say you have HA.

2000 users nam kubernetes danna wenawa. ow

 

[NEMISIS]

2000 users nam kubernetes danna wenawa. ow

ow, multi cloud redundancy ehemath dala ganna wei neh.

Spoiler: This

I am running some service on a Rasberry Pi, with my home internet connection, handling one request per about 10 second, that is about 9000 requests per day .

 

[ntayya]

ow, multi cloud redundancy ehemath dala ganna wei neh.

Spoiler: This

I am running some service on a Rasberry Pi, with my home internet connection, handling one request per about 10 second, that is about 9000 requests per day .

Ehema nemey ithin, boss kiwwama karanna oni ne api wada, mage wadak nam mama ochara wiyadam karanne na, Client ta salli prashnayak na

 

[NEMISIS]

Ehema nemey ithin, boss kiwwama karanna oni ne api wada, mage wadak nam mama ochara wiyadam karanne na, Client ta salli prashnayak na

Onam nam K8S gahala karanna puluwan, eth therumk na ban. sensible option ekak nemei. salli prashnayak na kiwuwata, denata $50 yana site ekata $500k witara enna gattama samaharu kamathi wenne na.
Godak un hitanne $50ta den run wena site eka, $100n HA karala, autoscaling dala ganna puuwan kiyala.

 

[ntayya]

Onam nam K8S gahala karanna puluwan, eth therumk na ban. sensible option ekak nemei. salli prashnayak na kiwuwata, denata $50 yana site ekata $500k witara enna gattama samaharu kamathi wenne na.
Godak un hitanne $50ta den run wena site eka, $100n HA karala, autoscaling dala ganna puuwan kiyala.

Ow math kiwwa, a unata mun ahanne na ne mama kiyana eka me welawe

 

[LZP1992]

Try AWS Lightsail