I can't imagine that every request would need to access all 16G every time. I use a method for caching data between requests using modules. I don't know if it would be truly shared between requests, or like Niphlod said, would be duplicated for each request. I have had a lot of success using modules + singleton pattern to hold small data in memory, but never used it with anything large. So maybe you should play with this idea and see how it responds.
- [web2py] keep shared objects permanently in RAM: Is it ... Sebastian E. Ovide
- Re: [web2py] keep shared objects permanently in RA... Bruno Rocha
- [web2py] Re: keep shared objects permanently i... Niphlod
- Re: [web2py] Re: keep shared objects perma... Sebastian E. Ovide
- [web2py] Re: keep shared objects perma... Niphlod
- Re: [web2py] Re: keep shared obje... Sebastian E. Ovide
- Re: [web2py] Re: keep shared objects p... Michele Comitini
- Re: [web2py] Re: keep shared obje... Ross Peoples
- Re: [web2py] Re: keep shared ... Sebastian E. Ovide
- Re: [web2py] Re: keep sha... Bruno Rocha
- Re: [web2py] Re: keep sha... Niphlod
- Re: [web2py] Re: keep sha... Michele Comitini
- Re: [web2py] Re: keep sha... Sebastian E. Ovide
- [web2py] Re: keep shared ... Niphlod
- Re: [web2py] Re: keep sha... Michele Comitini
- Re: [web2py] Re: keep sha... Mariano Reingart
- Re: [web2py] Re: keep sha... Sebastian E. Ovide
- Re: [web2py] Re: keep sha... Ross Peoples