Is Connection header mandatory in HTTP 1.1

I've troubleshoot recently an F5 load balancer issue and show a traffic like below (example with mangled fields)

10.178.155.180 - F5 
10.178.100.64  - server

Nr      Time        Source                Destination    Protocol flags  Length vlan hsrc                  hdst                  sport  dport  Info                           
    1 9.970000   10.178.155.180        10.178.100.64      TCP      0x02   100    777  F5Networ_45:c9:05    Rackspac_08:2d:3d     41588  80     41588 > http [SYN] Seq=0 Win...
    1 9.970821   10.178.100.64         10.178.155.180     TCP      0x12   100    777  Rackspac_08:2d:3d    F5Networ_45:c9:05     80     41588  http > 41588 [SYN, ACK] Seq=...
    1 9.971842   10.178.155.180        10.178.100.64      TCP      0x10   92     777  F5Networ_45:c9:05    Rackspac_08:2d:3d     41588  80     41588 > http [ACK] Seq=1 Ack...
    1 9.971921   10.178.155.180        10.178.100.64      HTTP     0x18   179    777  F5Networ_45:c9:05    Rackspac_08:2d:3d     41588  80     GET /status.aspx HTTP/1.1   ...
    1 9.978979   10.178.100.64         10.178.155.180     HTTP     0x18   475    777  Rackspac_08:2d:3d    F5Networ_45:c9:05     80     41588  HTTP/1.1 200 OK  (text/html)...
    1 9.979898   10.178.155.180        10.178.100.64      TCP      0x10   92     777  F5Networ_45:c9:05    Rackspac_08:2d:3d     41588  80     41588 > http [ACK] Seq=88 Ac...
    1 9.980028   10.178.155.180        10.178.100.64      TCP      0x11   92     777  F5Networ_45:c9:05    Rackspac_08:2d:3d     41588  80     41588 > http [FIN, ACK] Seq=...
    1 9.980315   10.178.100.64         10.178.155.180     TCP      0x11   92     777  Rackspac_08:2d:3d    F5Networ_45:c9:05     80     41588  http > 41588 [FIN, ACK] Seq=...
    1 9.980950   10.178.155.180        10.178.100.64      TCP      0x10   92     777  F5Networ_45:c9:05    Rackspac_08:2d:3d     41588  80     41588 > http [ACK] Seq=89 Ac...

The HTTP request and response:

GET /status.aspx HTTP/1.1
Host: www.dummy.com
User-Agent: MyAgent :)

HTTP/1.1 200 OK
Cache-Control: private
Content-Type: text/html; charset=utf-8
Server: Microsoft-IIS/7.5
Set-Cookie: ASP.NET_SessionId=tbc1vlvsjxjx01xs2frend5z; path=/; HttpOnly
Date: Mon, 25 Mar 2013 09:59:33 GMT
Content-Length: 9

Status OK

Problem

The 'Host' header is mandatory in HTTP 1.1. Is 'Connection: keep-alive|close' HTTP header mandatory in HTTP 1.1 as well?

Analisis and results description

An example request from the Chrome browser that specifically provide a value for the 'Connection' header:

GET / HTTP/1.1
Host: www.google.com
Connection: keep-alive
Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.22 (KHTML, like Gecko) Chrome/25.0.1364.172 Safari/537.22
Accept-Encoding: gzip,deflate,sdch

HTTP/1.1 302 Found
Date: Mon, 25 Mar 2013 10:15:23 GMT
Server: gws
Location: https://www.google.com/
Connection: Keep-Alive
Content-Type: text/html; charset=UTF-8
Cache-Control: private
Content-Length: 220

We can see that the client specify what should happen to the underlying TCP session. In this example the TCP session should remain active and idle for a while - until either the client or server decide to close it.

Although in our example we can see that the client (F5)

• didn't specify the 'Connection' header in the GET request
• the session was closed by the F5 (the client) after receiving the response 

According to 8.1 Persistent Connections  of the rfc 2616 the persistent connections are the default behaviour in HTTP 1.1. The field is not mandatory and either sides can close it. In our example the  client decide to close the established tcp session after receiving the HTTP response.

References

1. http://tools.ietf.org/html/rfc2616#section-8.1
2. http://www8.org/w8-papers/5c-protocols/key/key.html