Maybe it's just a throwback from Matrix 1.0 or 2.0, or perhaps my trying to apply some real-world detail to Matrix systems, but I have this tendency to want to divide hosts up into other systems that would control other facets of the company/corp, whatever.

After seeing some example, I realize that maybe having 1 Host system is enough to represent 99% of my needs, while using the "chokepoint" method seems to be the other 1% and at that, very rare.

Anyone else struggle with this or have other experiences with Matrix 3.0?

My problem is that I don't deal with enough dedicated deckers to get into complex Matrix topologies. Ah well, c'est la vie.

~J

you could easily create a single host that contains multiple virtual machines. that sounds like the best solution for what you're looking for. alternatively, you could create a full-fledged PLTG that contains an array of hosts (each of which might contain one or more virtual machines as well).

Another trick I've used alot in making Matrix runs harder for my players is when they Browse for something, come up with essentially a soft-link: that file is on this other host in the cluster. This forces them to 'Log onto Host' again, and it gives you (the GM) the ability to put more 'secure' things inside hosts that have higher ratings. Scenario is as follows:

Joe Shmoe is decking into ToyCo looking for the shipping data on the newest "My Pet Dunkie" dolls. The ToyCo host is relatively simple to find, and it scans as a Green-6 host, cake. Logging on, he finds Browses for the information, and finds that the production information is here, but on a different node. Logging onto the next node, he Analyses Host and finds that the rating here has all of a sudden jumped to Orange-6, an unpleasant suprise. Hopping out of that Orange sub-host, Joe finds that the personnel data is also located on that Orange host, and the security systems are located on a second sub-host. Guessing that the security host is probably Red, and crawling with IC, he decides to let the system cool off again, and come in later with a different load of programs.

This might be a skewed view of network security in 2065, but knowing how network security is in 2005, it makes sense to me. Perhaps that original 'host' should instead be a PLTG, but the concept is the same.

BaronJ

I seem to remember something in M3.0 about how a datafile could contain many virtual links that can lead the Decker through several systems before they actually find the data they are looking for. I'll need to do some re-reading.

I guess the hard part for me is wrapping me head around "a host" that could represent a large corp or some such. I like the PLTG idea, but (keeping in canon) these seem to be reserved for very large or government type operations.

As always, thanks for the input.

A PLTG is like a VPN today. A host is more like a LAN.

I thought PLTG was a private Telcom Grid? Like laying your own lines and stuff? I know WAN (PLTG) and VPN are similar in some respects, but I suppose I can treat them as you mentioned.

VPN is misleadingly used there, as its purpose is to create a virtual addition to a LAN. A PLTG is more like a private WAN, as Grinder mentions.

~J

I tend assume almost anything bigger than a small shop has a PLTG. I have a couple of "standard" chokepoints to sit behind my "standard" corporate front-end hosts that the public can access. I just describe different iconography and change up the trigger steps and IC layout for variety (or varying degrees of paranoia).

Even the high school the characters attend in my campaign has a public host and "secure" host for student records and such.

edit: Hmmm... I might be using the term PLTG wrong. Maybe I'm actually talking about tiered hosts.

Any place that runs its own PBX system is a good bet for a PLTG in 2060.

~J

Only really big (or expensive) places would need multiple hosts. Easiest is to front with a chokepoint, drop you main server behind it and hook up your others nodes to the non-security host. Its hyper expensive to run that many hosts but big business can obviously afford it.

A virtual host is a security trick to fake out deckers. It wouldn't work as a host very well because none of the data backs up. Its not real.

Another option to a single host is to bounce it. Its green 6 to allow the company to function at speed but when an alarm trips, bang, its orange 9. Adds challenge without stretching the bounds of reality.

i disagree--it says in Matrix that VMs can do everything real hosts can do. why you'd want to set up a VM, beyond decker-trapping and the testing of unstable code, i couldn't tell you. maybe it makes it easier to organize the hosts's functions and data. i'm not saying lots of companies use them for this, or anything--i'm just saying, if you want to get that ol' SR1 feel, VMs are one way to do it.

Each host you run has to have all the programs to make it opperate and takes a massive amount of memory. IC has HUGE multipliers and since your hosts will not be identical, you have to buy twice (or three etc.) as much stuff to make the host(s) work. A Green 6 and a Red 8 have nothing in common as far as programs go and you have to buy for each. They can be run off the same server but that just means that your hardware is singular. Since programs cost many times as much as hardware, you still pay a lot more.

As stated in Matrix, virtual hosts look and act like a real host. But, nothing you do to them affects the real host. Nothing saves, nothing can be editied. You are merely working in a little pocket host. Again, you have to pay for the IC, unless it is the same and the software to run it making it just as expensive as the regular host.

Virtual hosts are good for a lot of things. Companies contracting out their hosting will probably get virtual hosts on a larger machine. The big reason for multiple hosts over a monolithic host with virtual machines would be physical redundancy; if the single monolith goes down, you're screwed.

~J

i'm also not sure i agree that different hosts require completely different programs just because they're not identical. i'd be willing to bet, as a matter of fact, that software is one of the things that differentiates hosts--for instance, depending on the software you're running, the same physical computer might be configured as an orange-8 or a red-10.

I doubt it. I'll have to dig up my copy of Matrix, but there are some definite processing-power differences that come into play when you start programming on differently-rated hosts that I don't think make sense for the same physical computer without emulation/virtual hosting (which would only scale down).

~J

eh, depends. i mean, hosts of different color probably represent different hardware, generally, but i don't think differences in security value of up to, say, 4 points or so necessarily represent anything but more or less advanced security software.

Keep in mind that "host" object in SR is abstract enough that it could represent any number of computers, mainframes, smaller networks, etc. I try and keep that in mind when thinking of a multi-tier system.

I guess the question I should ask myself is, "Can I represent this 'host' system with a single set of numbers (AFCIS) or do I think it needs more tiers?". I know it's hard to throw off the real-life models where you count exact numbers of computers and think in abstract terms of SR.

I guess what it boils down to is does a multi-tier system add anything to my Run?

indeed: delicious, nutritious complexity.

I find that multi tiered systems are the way to go for any decking intensive runs. They allow the decker(s) to use all of their tricks and programs without the whole success or failure hinging on one roll. By using multiple host for a single instalation the GM may force light decking during the info search and planning phases (only need the green host to see if the target works there and in which office) and more complex decking during the insertion phase (the doors and cameras are on the security host) as well as, complications and twists to keep the runs interesting (the file and pay data are on a secure company server behind the choke point that you know from hacking the security server can be bypassed by phisicaly incerting the decker into the instalation.) In this way the dificulty can be more easily sculpted to your party's abilities and that pertinate information to the run is available to even less skilled deckers with out compromising the overall system integerity.

one thing about sr virtual macihnes, it seems that they dont run paralell but rather stacked. still one could theorise that one can run then in paralell, but there is no canon info about doing it so the effect it will have on the vm is hard to tell...

the host color seems to be a combo of computeing power and secuirty tightness. my guess here is that with higher security you have to add more prosessing power to handle all the extra logging and data prosessing as your basicly tracking and compareing more stuff then on lighter security.

basicly we are talking about a intrusion detection system (ids) that trys to figure out if the crashes and other stuff being reported all over the host is just random happenings or an attack. then when some treshold is reached it triggers of diffrent programs (ic) to investigate and counter the treat.

more data being gatherd means more accuracy but allso means more work for the cpu.

and depending on the host color it can be anything from a single box under the counter, running the mom n pop matrix shop, to any number of computers, working as a cluster and dealing with specialized tasks.

but then again the host is just a simulation done useing v2 of the matrix protocol, basicly its a abstraction to ease the workload of the users (need to have a frame do some backup after work? jack in and tell it what you need it to do like it was a normal person moveing physical stacks of papers and similar). multiple hosts may well be running on the same physical hardware. ibm and others have systems today that can run multiple os's on a shared cpu without useing virtual machines or other emulation-like tricks.