Providing Voice Recording Options
Cisco currently does not provide a VoIP voice recording solution, but rather relies
on technology partners to provide these types of integrated solutions.This section
will discuss the various VoIP solutions available.
One important point to remember is that VoIP voice recording differs from
traditional voice recording solutions. In a VoIP solution, we talk about bits and
bytes, as well as things like Real Time Protocol (RTP) streams, and H.323 and
Skinny Protocols.These types of terms require a completely different mindset.
Previously, it was fairly straightforward to add a voice recorder, since it basically
taped the conversation coming into the PBX. Unfortunately, we now have no
central point where we can place a voice recording solution, and even if we
could, it would be garbled because traditional solutions cannot understand and
decode IP streams (unless, of course, they were placed before the conversion to IP).
An easy way to think about this is to compare the following voice recording
scenarios. First, let’s look at a normal cassette recorder. Recording from a radio
station onto a cassette is easy. It’s a simple matter of placing a tape into a cassette
recorder, pressing record, and voilà, you’re recording from radio to tape.
Now, take the same solution, but apply it to an Internet radio station.
Questions arise.Where do you place the cassette recorder, and how exactly do
www.syngress.com
206 Chapter 7 • AVVID Applications
you record? You might respond,“I’ll put it next to my speakers and record there,”
but you’d still be relying on the PC’s ability to decode the IP packets coming in
from the Internet. Plus, sound quality would be poor.
The same goes for IP recording.While you can take a traditional solution and
crudely adapt it, vendors providing IP-based recording offer better solutions. But
how exactly is IP voice recording accomplished, and what should be taken into
account when looking to implement a voice recording solution?
As mentioned previously, it is possible to apply a traditional solution to voice
recording.This is achieved by placing the voice recording solution before any of the
conversations have been packetized, an option that allows all voice conversations
on a T-1 link to be recorded. But a problem arises in that there are no mappings as
to who the call actually went to.We may have a time the call recording started,
but given that we have hundreds of agents, we won’t know to whom the call was
transferred.Tracking down a particular call would be very laborious and would
probably require a client to listen to many recorded voice calls. As mentioned previously,
a good system will log all voice conversations, which essentially means you
will need a voice recording method with a large amount of storage space.
This is why a solution has been created specifically for IP telephony solutions,
and why specific requirements need to be met before implementing these
solutions. First of all, let’s discuss how to record these sessions, and how to decode
and provide Administrative information about the voice calls.
Current solutions rely on a feature available on the majority of catalyst
Ethernet switches, called Switched Port Analyzer (SPAN). SPAN mirrors traffic
coming in on one or more defined Ethernet switch ports and passes this information
to the SPAN port.This port is normally used in areas of network management
or when doing some type of packet decoding (which is done when we
utilize the VoIP voice recording solutions).The following is an example of how
the switch is configured:
Cat> (enable) set span 2/1 2/2
Enabled monitoring of Port 2/1 transmit/receive traffic by Port 2/2
Cat> (enable) show span
Destination : Port 2/2
Admin Source : Port 2/1
Oper Source : Port 2/1
Direction : transmit/receive
www.syngress.com
AVVID Applications • Chapter 7 207
Incoming Packets: disabled
Console> (enable)
To simplify matters, we need to put a VoIP recording application somewhere
where it can “hear” the conversations going on in the network around it. Because
the application receives all the IP packets, it can then make a decision (based on
configuration) as to whether the packets should be decoded, and where to store
them on the voice recording application server along with the necessary administration
information.While this is probably good enough for the majority of the
IPCCs, there are some limitations you should be aware of.
Firstly, you are limited by the speed of the SPAN Port. If it is only a 10 Mbps
port (normally it’s 100 Mbps), you may have problems sending all the information
to the single SPAN port.
Secondly, you must be able to configure a SPAN port (or something similar)
on your Ethernet switch.Without this, there is no non-intrusive way in which
you can monitor the voice traffic traveling your network.
Lastly, during a call, whenever ten seconds of silence occurs, the call normally
ends. However, if you were using an older version of CallManager (which did
not provide Music on Hold), and an agent put a call on hold while they consulted
another agent, , for every ten seconds the caller was on hold, a new call
would be generated. So, if the caller were put on hold for two minutes exactly,
we would actually end up with 14 voice recording calls (1 to start, plus 12 x 10
second intervals calls, and another to finish the conversation). Obviously, we
would not want this.To get around this problem, CallManager 3.1 and later do
provide Music on Hold, allowing you to have a single conversation voice
recording.
A workaround may be to utilize some of the capabilities of the catalyst
switches, which give you the ability to be able to configure your IP phone in one
virtual LAN (VLAN), and your PC in another VLAN.This would then allow you
to be able to SPAN a VLAN (as shown next), which in turn means the information
you are receiving via your SPAN port is only voice-related.
Console> (enable) set span 6 2/1
Enabled monitoring of VLAN 6 transmit/receive traffic by Port 2/1
Console> (enable) show span
Destination : Port 2/1
www.syngress.com
208 Chapter 7 • AVVID Applications
Admin Source : VLAN 6
Oper Source : Port 1/1-2
Direction : transmit/receive
Incoming Packets: disabled
Console> (enable)
There will be mechanisms whereby these types of voice recording solutions
will not have to use the SPAN port of an Ethernet switch, but will be agentinitiated.
These solutions will probably cater to H.323 capabilities, as well as silent
conference setups. Although we term it a conference call, in essence, there is only
the agent, the caller, and the third party, which is the voice recording solution.
When using the agent-initiated option, this is probably how the call would be set
up. Currently, two vendors are providing solutions that integrate with the Cisco
IPCC:
Nice www.nice.com
Eyretel www.eyretel.com
For more information on up-to-date voice recording solutions as well as
other IPCC partners, visit www.cisco.com/warp/public/180/prod_plat/
cust_cont/ipcc/part_doc.html