Split a wireshark capture in a smaller file with editcap

One of my customer sent me a tcpdump trace with a size of 2.5 GB. It goes without saying that it’s too large for open it on a desktop PC with wireshark.
So I discovered the tool editcap delivered with wireshark which allowed me to split the trace in a smaller file. Hopefully, my customer gave me the exact time where the problem has been encountered so I can get the only traces related to this event.

C:\Program Files\Wireshark>editcap
Editcap 1.8.7 (SVN Rev 49382 from /trunk-1.8)
Edit and/or translate the format of capture files.
See http://www.wireshark.org for more information.

Usage: editcap [options] ... <infile> <outfile> [ <packet#>[-<packet#>] ... ]

<infile> and <outfile> must both be present.
A single packet or a range of packets can be selected.

Packet selection:
  -r                     keep the selected packets; default is to delete them.
  -A <start time>        only output packets whose timestamp is after (or equal
                         to) the given time (format as YYYY-MM-DD hh:mm:ss).
  -B <stop time>         only output packets whose timestamp is before the
                         given time (format as YYYY-MM-DD hh:mm:ss).

Duplicate packet removal:
  -d                     remove packet if duplicate (window == 5).
  -D <dup window>        remove packet if duplicate; configurable <dup window>
                         Valid <dup window> values are 0 to 1000000.
                         NOTE: A <dup window> of 0 with -v (verbose option) is
                         useful to print MD5 hashes.
  -w <dup time window>   remove packet if duplicate packet is found EQUAL TO OR
                         LESS THAN <dup time window> prior to current packet.
                         A <dup time window> is specified in relative seconds
                         (e.g. 0.000001).

           NOTE: The use of the 'Duplicate packet removal' options with
           other editcap options except -v may not always work as expected.
           Specifically the -r, -t or -S options will very likely NOT have the
           desired effect if combined with the -d, -D or -w.

Packet manipulation:
  -s <snaplen>           truncate each packet to max. <snaplen> bytes of data.
  -C <choplen>           chop each packet by <choplen> bytes. Positive values
                         chop at the packet beginning, negative values at the
                         packet end.
  -t <time adjustment>   adjust the timestamp of each packet;
                         <time adjustment> is in relative seconds (e.g. -0.5).
  -S <strict adjustment> adjust timestamp of packets if necessary to insure
                         strict chronological increasing order. The <strict
                         adjustment> is specified in relative seconds with
                         values of 0 or 0.000001 being the most reasonable.
                         A negative adjustment value will modify timestamps so
                         that each packet's delta time is the absolute value
                         of the adjustment specified. A value of -0 will set
                         all packets to the timestamp of the first packet.
  -E <error probability> set the probability (between 0.0 and 1.0 incl.)
                         that a particular packet byte will be randomly changed.

Output File(s):
  -c <packets per file>  split the packet output to different files
                         based on uniform packet counts
                         with a maximum of <packets per file> each.
  -i <seconds per file>  split the packet output to different files
                         based on uniform time intervals
                         with a maximum of <seconds per file> each.
  -F <capture type>      set the output file type; default is pcapng.
                         an empty "-F" option will list the file types.
  -T <encap type>        set the output file encapsulation type;
                         default is the same as the input file.
                         an empty "-T" option will list the encapsulation types.

  -h                     display this help and exit.
  -v                     verbose output.
                         If -v is used with any of the 'Duplicate Packet
                         Removal' options (-d, -D or -w) then Packet lengths
                         and MD5 hashes are printed to standard-out.

So, the following command allows you to get the only part of time of interest :

-A : specifies the start time
-B : specifies the stop time
capture2.pcap is the input file
capturetest.pcap is the output file

C:\Program Files\Wireshark>editcap -A "2013-08-26 15:20:00" -B "2013-08-26 16:00:00" "D:\Users\capture\capture2.pcap" capturetest.pcap