Whilst the above is probably true, it no longer holds, most processors nowadays are at least dual-core, in other words, we can now do TWO things at a time!
Now, the above is the newest argument I’ve seen against VoIP systems such as asterisk! Can you imagine it?
Ok, so to quote the woman:
Attached you will find a copy of my quote , I have done quite a bit of homework on ULS and my feedback on it although I cannot completely back this up As I probably have to do a lot more homework is , that yes companies have tried it Shell for example and it has not worked for them. I am still speaking to people , and can get you names. From what I understand , A pc works differently purely because it can only accept one command at a time, whilst a PBX can handle a lot more. This results in dropping off call so forth. Please look through my Quote and if you have any questions please call me.
Ok, I’ve got to admit, when I first read that I was like flabbergasted. I did not know how to respond to that. It’s actually pretty funny. Woman, firstly, I seriously doubt you did actually do quite a bit of homework on ULS. Heck, you’re still in an era where you believe a PC can only do one thing at a time. Grow up and smell the 21st century. With the advances in CPUs alone there are now systems out there that can really do 100s of tasks at the same time. Some of these systems are in a single box. There’s been patches that went into the mainline Linux kernel about two or three years ago that allowed it to scale quite horrifically in terms of number of CPUs (1024+ kind of thing) with insane amounts of RAM.
Ok, so let’s look at a switch that handles a few hundred calls a day at a call concurrency of up to about 15 concurrent calls at any given point in time, it’s a 8-way SMP machine, 2.5GHz Xeon (quad core, so two CPUs actually), with 2GB of RAM. It’s idling at an average of 3% of the CPU with a max load of 9%. No, that’s not 9% overall over all the CPUs. It’s 9 % of ONE of the CPUs (yes, assuming near linear scaling this will thus imply a max call concurrency of around 1200 calls on a SINGLE MACHINE). Physical RAM (including disk buffers and disk caches) is cruising at just under 600MB actually committed to stuff!
Also, consider that packetisation takes place AT THE PHONES or on the DIGIUM cards already, so all asterisk is really doing is packing and unpacking voice fragments into udp frames and passing it on. It may, or may not, copy these packets to disk for voice recording purposes, and it may transcode these fragments from one codec to another. Sounds CPU intensive? You bet! So how much CPU do we need? Not a lot, actually, I’ve got P4 (2.4GHz) systems out there performing full transcoding and recording on all calls that are using no more than around 15 % of it’s CPU with a load of up to 8 concurrent calls. Sounds hectic? My mail server is pushing (without spamassassin enabled) load averages of 2.8+. That’s hectic. A 15% loaded CPU would have a load average of approximately 0.15. That means there are still cycles to spare. In fact, it means quite simply that 85 % of the time the CPU is actually waiting for instructions to execute!
Ok, so what do you say to these kind of lunatics that believe that their PBX systems are still better than a software based PBX? It definitely has it’s advantages, lower latency, echo isn’t as serious an issue due to the lower latency and the handsets can usually take care of it. Also, much less hybrid balancing required because … there is no hybrid in the PBX and the phones needs to handle this. Oh, and it’s obviously longer in the industry, so it’s “better tested”. I can’t think of ANY other advantage right now.
Yea, just go read Asterisk – The Future of Telephony, specifically chapter 7. Then come back with some serious arguments. Dropped calls etc are not due to the PC or to asterisk (99.9% of the time) but rather due to a bad configuration generally. I didn’t have any dropped calls in my office that I couldn’t trace to misconfiguration (mostly due to me messing around with the PBX to try new things) any time in the last six months. Now, in terms of stability (which is also something people have attacked in the past):
System uptime: 6 weeks, 4 days, 28 minutes, 2 seconds
That was when I last rebooted the machine on request from a support technician at Internet Solutions, where the particular server is hosted.
Ok, I better end this rant right here right now before I start throwing people with phones, But since you want names of big companies that’s making use of VoIP … let’s mention that at least one of the major banks in South Africa is making use of VoIP. Let’s mention that Internet Solutions is making use of VoIP. Let’s mention that GSM is in fact nothing other than VoIP, just with an alternative network level carier (ie, GSM instead of IP). Let’s wake up and see that even 5 years back the push has been to get IP down directly onto the fibre and eliminate SONET and ATM, enabling ALL data (yes, voice is simply data) to be carried between any two points as part of the network. H323 has long since run on IP.
Oh, and what if web servers could only serve one request at a time? What if you had to use either your web browser OR your mail client like back in the days of DOS? Woman, have you ever used a PC any time after the days of DOS/Windows 3.11?
I would have mentioned the advantages of VoIP (and asterisk specifically) at this point, but there are plenty other resources out there.
uhm… maybe she is one of those hoe lived in a hole for 40 years because they were afraid that the cold war will bring a nuclear holocaust… the end of earth as we know it.