Does the poor performance correlate with the spikes? If so, your could argue that the key is noticeable to a human visitor and so worth asking the host about. (With so many metrics these days, it's easy to forget to ask whether what we're measuring is something an actual person would notice).
That said, it can be hard to have enough data to be sure there's an issue. For instance, sometimes a slower response time is entirely down to poor conditions on the route between the monitoring server and the web server. A different host wouldn't travel over the same carriers so wouldn't hit the same problem.
I think I'd ask the host to look into it, giving them exact timeframes when things were slow. They may not find anything, and that may not reflect badly on them if they can't, but equally they may.
This followups up on a series for 4 to 5 minute sessions of lost connectivity. Once or twice was no big deal, but it was happening several times each week. When I asked, I was told there were no problems. I did provide the times and such.
I felt as if I got a bit of a brush off. Again it's not big deal, but since I'm testing between two comparable services this might be the decider.
Contact them immediately when you notice high response time. Also provide the tracert report from your local system. It may possible that the server is overloaded hence the response time is high or there may be little network issue.
High response time is not good and it also affects SEO for your website.
What service are you using to test the server response time?
I would suggest you to use pingdom.com with a free account (not tools.pingdom.com) and see what the average response time is over a period of say one month or so. Pingdom will check the site every 5 minutes using different servers located around the world. You can then log in to your account and check the response times for each location. That way you have better chances to identify what the problem is. You should expect a higher response time for servers located in a different country than your host's.
450 to 600ms is not that slow. the question is what kind of DOM rendering is going on server side.
With Nginx and Varnish you might be able to shave the response time on a really well coded piece of PHP down to 250ms-300ms, at the cost of compatibility (unacceptable for shared hosting). But even with an extremely optimized server architecture on a dedicated server and a 1Gbit connection, a heavy PHP page is going to take minimum 400ms in response time.
For shared hosting those response times are not bad at all. Now something like an HTML page, you'd expect an 80-100ms response time. New Relic can be a useful tool in figuring out where your load time bottlenecks are.
██ BitronicTech Affordable Hosting and VPS Since 2007. Check out my Blog.
██ 99.99% Uptime Guarantee. 30 Day Money-Back Guarantee. 24/7/365 Support.
██ cPanel/WHM - CloudLinux - Varnish - CloudFlare - Softaculous - Daily Shared Hosting Backups