ISP looking to add HTTP caching server

Hey there! Apologies if this would be better placed in sysadmin, but thought I might try here first.

I work Network Operations at a smaller ISP. We recently implemented both Netflix and Google caching programs with great results. Results were so good that my superiors would like us to implement even more caching. With Netflix and GGC already implemented, I feel like we are starting to scrape the bottom of the barrel.

We've started researching into some commercial caching products, but few seemed geared towards the ISP. Most are geared towards the enterprise, who of course have far more control over their users, and can implement proxy-pacs and the like. We've seen a couple products that may work for us, but many of them require adding a significant amount of complexity (i.e. PBR). I'm apprehensive to implement PBR as it puts a lot of technical debt into the network, not to mention that a majority of our Network Operations department would have difficulty troubleshooting an issue with it.

Anyways.....

Some details:

• 25-30K subs. Mostly residential. Some college dorms/apartments. Some very large businesses

• Peak edge/internet traffic is around 40gbps (including already cached traffic).

• Our uncached-HTTP traffic is about 6-7gbps (this excludes SSL/TLS traffic)

• Core and data-center network is exclusively Cisco XR, NX-OS, and XE.

• Many 10G peerings to the internet, including one IXP. Looking to add another 10G IXP next year.

• Superiors are pushing hard for this for cost-savings purposes, not necessarily performance.

• XE and XR equipment does not support WCCP

My questions are:

• Do ISPs still implement generic HTTP caching servers in this day and age?

• What solutions would you recommend for this (Open Source, Commercial, etc...)

• What other caching solutions should we explore (Akamai, for example, although I hear their requirements are high)

• Is it even advisable to implement a generic caching solution when Netflix Caching and GGC are in place? (Possible negative interactions)

• How much HTTP content can we reasonably expect to cache? Or rather, how efficient could we be with this?

• What other pros/cons are there with a solution like this that I'm not thinking about?

• Alternatives?

Thanks all for any assistance. Let me know if there is any additional info I can provide that would help.