Recently, I started working part-time in server construction, and during the training, I was asked to make suggestions on server configuration to run 1 million PV media sites per month.Media site settings are as follows:
·1 million PVs per month
·It is irregularly covered by Yahoo, Smart News, Gnostic, etc.
·I am currently renting one Sakura Internet business professional.
·Sometimes the web page is not displayed when access is concentrated.
·The person in charge of the web at the media site management company is an amateur when it comes to servers.On-premises is not possible.
·Operated using Wordpress in a LAMP environment.
·The size per page is about 1MB
• Access is concentrated for a total of 6 hours, day and night.
I'm thinking of moving to a cloud server.
How should I determine the CPU and memory configuration at that time?
I think experience speaks for sizing, so if you have any information about what kind of app it is and what kind of configuration it is, you may get answers from people with knowledge.
If it's about 1 million PV per month on Wordpress' blog site, I think it's about 2 cores and 4GB, but it depends on the app and content you post, so I can't say for sure.
Also, if you're in the cloud, CPU and memory configurations will be a choice from the usual pattern, but it's easy to change, so if you don't know the best size, it's more realistic to look for the best configuration while checking load conditions for the first time.
This is because even if you estimate the server resources you need in advance, they are virtual machines after all, so they may not perform as expected (although they can be dedicated to H/W if you pay for them), so I think it's essential to operate with system scalability in mind.
If the person in charge of the web is an amateur, it would be helpful to automate it as much as possible.For example, by separating the Web from the DB, the lack of Web performance is automatically scaled out and resolved.It might be a good idea to notify DB of insufficient performance through alarms and scale up.
Other respondents say they need more detailed information such as requirements and load per PV.
One thing I can say is that the introduction of a load balancer is essential as an element that is not mentioned in the question.For example, AWS is Elastic Load Balancing.
• Unscheduled coverage by Yahoo, Smart News, Gnostic, etc.
Therefore, I think there is a limit to increasing the specification of a single server, and in the end, when access is concentrated, the problem of not being able to view the page cannot be solved.
Since the number of servers increases and decreases without changing the appearance of URLs from users, it is safe to have at least two servers behind it, not just cloud on-play.(Because it is not uncommon for servers to go down without increasing access)
© 2024 OneMinuteCode. All rights reserved.