What is the difference between a hacker's comlink and the Corporate Node he's trying to hack? The node (what used to be a "host" in SR3) has the same system/Responce rules and AFIK all of the host's atributes are caped at 6 right? So how does a corporate host handle hundreds of users and bunches of IC without the maximum number of subscribers exceed the system rating, or having so many agents/progs loaded and running that the system is overloaded to the point that the responce=zero and the whole thing crashes?

If a comlink and Host are not the same thing then what are the rules for designing a host and why the hell aren't they realeasing the next matrix book right f.kn now?

edit: i have searced the board and I swear I saw a post about loading agents with agents that self spawn until the system crashes but I can't find that thread

Thanx

Currently, a commlink works the same as a corporate node, with the same limitations and all, subject to GM common sense.

OK, since my sense is common to < 2007, and I've never met anyone from the shadowrun universe in 2070, could you spell it out for the retard (me) please.

If 500 researchers in the corperate liberary load a rating 1 agent + Browse then go to lunch, does the system crash?

If not, is there any limit to the numbers of attack IC's I can launch in responce to a system alert.

If there is no limit why can't my hacker crash any system with the self spawning agent thingy.

If there are limits of somekind how do more than 11 agents access a node at one time?

Normaly I'd adjudicate according to my RL knowledge but I'm pretty much limited to porn surfing and photoshop and PHOTOSHOPING PORN! so I realy have no idea about what might be right.

thanx

I work it this way: You may only subscribe to a limited number of nodes, but an unlimited number may be subscribed to you. It all depends on where the icon is. If your icon is on their node, you're subscribed to them. If their icon is on your comlink or node, they're subscribed to you. Not necessarily right, but it makes sense in my mind.

[edit]Agents running on a node is a whole other ball of wax. I don't know a defense to that.[/edit]

No I kind of get it, programs I'm running on my comlink; analyze/browse/edit, don't lag the host node because I'm using my processing power to run them, If I unload my agent from my persona to the node the agent is running on their system so it uses their processing power not mine I just can't immagine any student / wage slave not having their own rating 1 agent loaded with browse. Or if i realy don't want to do any work today a rating 2 agent loaded with browse and edit.

"Boss I'm taking a long lunch, I'll have that report to you by 12:30 though."

Ah Ha! I think I just realized what a "network" is. I don't load the agent onto the company system, i still run the agent on my comlink subscribed to the company host while I do something else. "Marge I've just tripled my productivity" (homer simpson)

The only programs running on a host node are IC and DataBombs since any users or "utility" agents should be running on their own node NETWORKED to the host.

Sweet! so when designing systems for players to hack. I can trigger tonns of attack IC for an alert. By the rules it doesn't matter if the responce is seriously degraded since every agent will get 3 IPs, and they all get a turn at the gangbang every pass.

Heh; Grey IC, Black IC, GANGBANG IC

My SGM (see signature) tries to answer some/all of these questions. Though, I have to say, just as a workarround.

The agents issue was discussed extensively here:
http://forums.dumpshock.com/index.php?show...=15007&hl=agent

I like the idea that in research facilities, libraries, and the like that there would be a bank of provided rating 1 agents on a rotating que to help visitors. You would send it queries and wait for it to do the leg work. Kind of like rare-books libraries work in the meat world - only with Agents instead of Grad Assistants. The agents would run from thier own systems which would keep the bog-down from happening on the main system. You could have your agent send the requests from your end, but that's your end. It would be kind of fun to see what kind of 'telephone game' mistakes having a search run through two low rating agents would make.

Of course, with data size being such a non issue, and comms being so cheap.... visiting a major online library could be more like visiting one mid rating commlink out of a bank of 100s of seperate commlinks with identical information on them than visiting a more centralized system.

@ Serbitar

Thanks for SGM, scanning it now. I didn't notice it when I was looking thru your cyberlimb house rules. (btw, thanks also for the CyberStuff, not quite the flavor I want in my campaign but still cool)

And again thanks for that link to the other thread, I had read that one last night befor I started this thread but that wasn't quite what I remembered reading a couple weeks ago.

@ The Jopp
What I'm thinking about GANGBANG IC isn't so much about hosting the IC agents on another node but the idea that boging down the system isn't such a bad thing. Here's an example of what I'm thinking, (bear in mind that I just got home and I havn't read thru Serbitars stuff yet)

Corporate Host Node: Research Node.

Node Rating 4,
networked on a wired LAN in one building, access is passcode protected.

SUBSCRIBES TO:
3 other nodes on this LAN;
... Acounting,
... operations,
... Management.
RTG
PLTG

CURRENTLY RUNNING PROGRAMS:
IC. Trafic agent 4/ Attack 4, Analyze 4, Stealth 4 (when unwired comes out the attack will be tarbaby) [note: as i read the rules programs running in an agent do not count against the node's OS for responce]

This IC agent is hidden in the node, (Stealth 4) it constantly takes Observe In Detail actions, (six times every turn), scaning the node until it notices an agent, user, program, file that it doesn't recognize.

Once that happens this IC will take another Observe in Detalil action, Scanning the suspicious icon. If the suspicion is verified the Trafic Agent will trigger an active alert.

ACTIVE ALERT:
An active alert runns several Additional IC and calls a Security Hacker to the node.

Alert turn 1, IC programs launched.
Alert IC: (these agents are directed to the intruder by the Trafic Agent)
1 - tracking agent 3, /armor 3, Track 3, Stealth 3 (begins to track if intruder present)
2 - Attack Agent 3 / Armor 3, Attack 3, Blackout 3. (begins to attack if intruder present)
1 - Cleric Agent 3 / Analyze 3, Browse 3, Medic 3 (begins to analyze system, files, programs, personas for damage and repair.)

There are now 5 programs running; Node Responce reduced by 1 wich reduces system to 3 wich reduces ratings of all agent and programs of agents to 3

Alert turn 2, (If alert conditions are still present.)
Alert IC:
3 - Attack Agent 3 / as above.

there are now 8 programs running, node responce reduced by 2. ratings of system/agents/programs reduced to 2. Security hacker called.

Alert turn 3, (If alert conditions are still present.)
Alert IC.
3 more attack agents rating 2


There are now 11 programs running Node Responce is 2, agent and program ratings are 2, but the intruder is being attacked by 8 rating 2 agents, and tracked by an agent he may or may not have noticed.

I probably wouldn't do it exactly this way, If I had been thinking I would have stepped things so that responce remained higher a little longer but I think you can see what I'm getting at.

I can't believe we are visiting this discussion again. can't people wait till Emergence comes out for the expanded rules. All your questions will be answered then. In the mean time think about the practicality that a Corp host that has millions of people view every second being run off even the most powerful Commlink. It is loco to think that. A Corp host is either a lot of Commlinks together with a shared database or a supper computer that doesn't follow standard rules (and the expanded rules will come out in Emergence).

Okay so let's come up with something.

The assumption that I make is that they have a fairly high response 5+ but they have so much computing resources that the response never lowers. This way we can keep the numbers from being absurd without the overload of response. You can consider this a special attribute of servers.

Signal is not an issue so ignore that.

System would be high to accomidate everything but also ignore program limits.

Firewall would have to stay in a reasonable range so hacking would be possible so keep it in the 1-6 range for everything except highly secure meagcorp or military servers and they can go to an 8+.

I don't think that they would have 15 IC or Agents running around the server. They would just spawn a new copy for help requests and during security alarms.

Set a hacking threshold and have that as a security level (say 1-4). Anytime they roll below that threshold then for every point they go below it they add a point of alarm. Also accessing certain files or nodes could raise the alarm level until it is confirmed.

Drop the perception check by the server checking for hackers. Just make it passive because if it was an active system then They could just sit around making perception tests with a near infinite number of IC and always win. This way the hacker has a chance. Also make their stealth the alarm's threshold to activate (ie you have to mess up enough to tip off the server but you can cover up a certain amount of mistakes).

That is what I suggest. I use a different version of the matrix system but this would be close to the original version and should work.

Thanx Garowolf, we're thinking along the same lines.

At lunch today I had a brainstorm about "distrubuted processing" which is a phrase that I feel justified in throwing around because I'm completely computor illiterate. (and if you check my spelling you'll notice that I am almost completely engrish illitelate. I blame my nation of origen.)

anyhow, back to my "distributed processing" which might be something like several computers working on the same task.

Take several computers, hardwire them together, all three or four of them use rules similar to "Teamowork Tests" pp. 59 to run the OS.

I havn't had time tonight to develop this much but basicly each additional unit provides a bonus to responce. The point of rules being to establish limits I'd cap this a 5-7 total units, but I have a vague feeling that today someone is working on linking hundreds of processors to work at the same time re. Mac's Quadcore G5 desktop and Dell's dual core2duo machine but... Bigger?

So maybe you can have a core unit with additional units up to the rating of either the responce of the core unit or the system rating of the OS.

I get the feeling that the intention of the developers is a device rating cap of 6. So max system/program/responce/firewall/pilot/skill/etc. rating should probably be left at 6. and the mechanics should reflect not a rating but a bonus.

something like this;
core processor rating 6
system rating 6
6 additional processing units

each additional processing unit, a device unto itself, adds 1 to the "EFFECTIVE" responce of the NODE. All of the devices together form a single node. ACTUAL responce rating is still 6 for all purposes relating to tests i.e. the Initiative of a security decker or the maximum rating of a Host's verson of the Reality Filter program used to skulpt Ultraviolet nodes.

Each program/agent run on the distributed node counts against the threshold for responce degredation. HOWEVER, (hehe I like caps), However the "EFFECTIVE" responce of this theoretical system is 12. [core rating 6 (+1 x 6 extra devices.)] SO, every 6 programs active on the node reduce "EFFECTIVE" responce by 1 and ACTUAL responce is unaffected until you have 36 progs running. Which seems to me like overkill and waaay to powerfull but I'm of the opinion that programs loaded into agents DON'T count against the system responce threshold.

Other ideas off the hip;
crashing the node: any of the linked devices is capable of running the the whole node at their individual device ratings. Crashing one of the computers (I can't realy think of them as comlinks and at this point they aren't nodes of themselves.), will lower the system and responce rating but you'd have to crach all six to kill the node.

double OS cost: no normal OS progs can run a distributed node these are special versions of stock OSsesz like Mac's OSX server. or the windows equivelant. (I should disclaim that I have any idea that I know what RL server software is or does I only know that it's special and costs more.) So the Isis Orb OS prog would cost 2000¥ or more and have the same stats.

Responce limited System rating: One of the special things about the distributed node system software would be that it's not limited by ACTUAL responce. So you may be able to run a rating 6 system OS on a distrubuted node with rating 4 responce as long as the EFFECTIVE responce is at least 6

I'm starting to get punchy, going to bed now. would love to work on this with people who actualy know somthing about electronics and/or the SR4 rules.

Grinder

I didn't read that Garrowolf was throwing out the hacking rules. I think he, like me wants his hacker player to be able to hack something that isn't a comlink.

Please, how do you set up your Corp Nodes in your game?

Everyone, what are the nodes you've run against? Which ones kicked your ass which ones where easy, did they all seem just like hacking a high level target's comlink? What made them different?

ok now I'm really going to bed, you bastards are wearing me out

well to be fair to Grinder I did change the hacking rules significantly. These server rules would be farther afield in my game but mostly how I put it here.

In my game I removed the limits on Response unless you are running more then one agent and then it drops alot. Servers wouldn't have that issue.

I also turned System into a package deal of software like OSs are now and removed the system limit on programs so response is the only limit. That way you don't have two traits doing the same thing (limiters).

It's just an alternate system. I never consider cannon to be anything more then the best they could come up with without me!

I like the idea of distributed processing too. The way I tend to see it, is that you don't have a big corporate server, but a lot of smaller nodes merged together. When you connect to a corporate server, you connect to one node, reserved for you.

Thanks to a good use of distributed processing, you don't see that you're in only one node : you can freely go from one node to another without even seeing that they are two different nodes.

For example, a security server is linked to 9 surveillance cameras. The server consists of 3 rating 5 nodes (security terminal). Each one is suscribed to 3 cameras and to the other nodes. They have 4 slots left for suscribing to user's commlinks.
Each of them run programs and agents (the agents can roam in the whole server). From node A you can check a camera suscribed to node B : node B just sends the data to node A. So the host can look like a monitoring room with 9 screens.

This way, you don't have to use different rulesets for each kind of nodes while still being "realistic".

A "redundant array of independend Comlinks" as a server model... nice. Could even be hardwired for the most part and would still be mobile.
What if your hacker does this? (Any program at rating 6 and running, yes!)
The number of subsriptions will hardly be an issue - larger bandwith for servers will be possible. I do even think the subscription limit is for comlinks only.

I do not think that the multiple-ICE-approach to intruder repelling should be "forbidden". It should be the method reserved for ultra-secure hosts, as the licensing costs are surely high - but thats one way to provide security to Z-O.



So far, we basically set things up by the book and do matrix security by storytelling needs. Encryption comes with an agent that constantly re-encrypts (exchange the extended test with a normal test, now it works. while staying RAW). Most simple systems only use the admin user-level and have alarms in case of user registration. We consider to allow increased ratings for stationary systems. Basic reservation against that idea stems from the necessary changes to vehicle ratings.

Personally, I just don't worry about loads, agents or IC when it comes to corporate nodes...the hacker still has to worry about it because even though he is in the Matrix, his own commlink is his connecting point to it.

So, I basically use all the same rules when hacking a node, minus the reduced response, system, etc. This could be abused, but as a GM I try to keep things fair and it hasn't become a problem.

I forgot to mention, but even if I take a "node array" approach, I consider different "node arrays" for each section of a corporate host, just like GrinderTheTroll.

I tend to view hosts as a building, with "node arrays" as the rooms. Everything is far simpler in the Matrix as soon as you start thinking by comparison to the physical world. Creating large-scale system isn't really hard, but as for creating a large building, it takes some time to do it properly.

I have no problem with corporate host using the same rating scale because corporate host run on a far larger scale. You're the only one using you Edit program on your commlink to write your e-mail. There are thousands of people doing it on the corporate node. Most of the times, there's only your own persona on your commlink, there are thousands on the coporate node...

The corporate host is far more powerful. It's just that, divided by the users and the traffic, it's as if there was only you and your traffic on a single commlink node. This way, you don't have to use two different rulesets or scales, you're still able as a hacker to hack a regular corporate host but you can still have difficulties hacking a secure commlink : it's balanced, streamlined, and "realistic".

Nothing wrong with that if you play it right and don't resolve every matrix run this way (as you don't resolve every social encounter by roleplaying the entire conversations if your players start asking people on the street about something)

from Blade

Just a reminder, there is nothing saying a limited number of Personas can access a node. Personas don't cause node decay, unloaded IC/Agents with offline Personas so. The problem arises when Personas/Nodes unload an Agent into the Node and take their Persona offline.

In the case of a Node/System, they can't take themselves offline, so it still suffers from having too many IC/Agents loaded since they must draw resources from someplace.

Right. Thanks for pointing that out, I used to think that a node had to subscribe to another one to be linked to it.
But that may be the case as well : cameras and door don't have personas, but they logically have a subscription list.

Anyway, I think the best is not to bother with that, and just consider these two points :
1. Corporate hosts can be linked to a lot of nodes and your commlink can't.
2. Due to the massive traffic they have to deal with, corporate hosts ratings and rules are the same as commlink rules, even if they are far more powerful.

This way you have game balance, streamlined gameplay, canon rules and "realism". Just consider the "node arrays" as a way of explaining why the rules are this way.

So you connect your commlink (A) to the node-array (B). You ask (B) for Combat-Agent. It sends along the Agent and the Agent then draws resources from its host node in the array yes? So you could have lots and lots of Agents in your Target node, all drawing Reponse from their host nodes (since they are still online?) instead of your Persona itself?

Is this what you are saying?

There's a simple way to work around this : just consider that the persona is linked to a single commlink, even if this commlink is part of an array. The persona can't load programs from other nodes than its own and can't run programs other than programs it carries or programs on the node.

So, the persona can only carry the programs from his own single node, and can only run them on his own commlink.

The "node array" is quite simple really. Here is the idea :

1. It's strange that your big corp system isn't more powerful than your commlink (same ratings). They have enough space to store a lot of big computers, so why won't these computers be more powerful than commlink ?
2. The use of the same rating for commlinks and corp system is quite good : you use the same rules for hacking one or another, and you can have as much trouble hacking both. If there was a big differences between the two, the hackers would be able to hack commlinks far too easily, or wouldn't be able to hack corps system.

So I tried to find a way to get both 1 and 2.

What's the difference between your corporate server and your commlink ? The traffic. Your commlink deals with far less traffic than your corporate server. That's why I thought of viewing the server as "node arrays". Basically, it's just a lot of smaller nodes, so that each nodes deals with as much traffic as your commlink. So your corporate server isn't more powerful, it's just bigger. For this to work, you just have to group the node so that they look like one single big node.

Basically, it was just a RP justification of the rules. But OneTrikPony's remarks brought up some interesting issues, and I tried to answer them. (and maybe I didn't do it properly).

--

Agents can hack nodes, but can they load themselves on another node to exploit it ? [Edit] Looked it up, looks like they can. But then again, as they are loaded on the node, they can be unloaded from the node by anybody with an admin access. [/Edit]

I totally agree with you.

I think that the use of nanotech and things like that made that size doesn't limit the power of the devices anymore.
But you can get a lot more computing power with distributed processing, that's why I thought of arrays of smaller machines. This is also far better for the stability of your server (if one node crashes, the other can still manage) and for the flexibility (if you need more power, you just have to add some devices).

For more powerful machines, I'll consider high ratings (up to 8 for the best of the best), complicated systems (with, for example, a chokepoint node/node-array, and different nodes/node-arrays for each part of the system).
To keep that out of player's hand, you have the cost of all the devices and the size needed to store them all. A player can get two commlinks and use one as a security chokepoint between his "real" commlink and the outside world, but 30 commlinks would be a bit bulky (and expensive if you want high ratings).

The problem is that if you start adding in the array stuff you end up with an end run around a rule that we do't really need. Just drop the response and system limits in the first place and you don't need to force your players to pay more to counter act it.

This would also solve your problem with your servers in the first place.

They already said that memory is not an issue so why worry about it. Besides if we really want to get this technical about response and memory levels and processing power then why does a scan program that is a minor function of what Windows does today slow down a commlink as much as an increadibly power Agent that is beyond what we can do today? Just because they are both programs doesn't mean that they would have the same resource level access. Just drop the response stuff as a limit for the number of programs and leave it as a limiter of rating.

Exactly.
And that's why (I don't want to bother everyone by constantly hammering this, but it seems that I wasn't clear enough about it) I'm talking about node arrays. The point is not to add rules, complexity or anything. It's just to tell your players that it's this way because :
"Due to miniaturisation, the Corp servers can't be much more powerful than your commlinks. Of course, they can get a lot of small computers together to form a big mainframe, but it won't be more powerful, it'll just be able to deal with far more trafic."

If you're incomfortable with this and if you're players aren't the kind who argue each time they find something illogical, just forget the whole idea and consider that "due to the way things are, commlinks can be considered as powerful as corps server."

military organisations and such can have higher than rating 6 commlinks.

AAA corporations have militaries.

therefore, AAA corporations can have higher than rating 6 commlinks.

therefore, AAA corporations are not limited to the same hardware as john Q public... however, for most uses, they aren't going to bother setting up a rating 7+ system, because it isn't generally needed. this isn't to say that they don't have access to it, but rather that it isn't efficient.

for example, AAA corps have access to laser weaponry, i think it's safe to say. this does not mean that security goon A at the weapons world store down at the mall has it... but if, on the other hand, you are trying to sneak into a top security research lab to steal a prototype or something, you might run into laser weaponry.

generally speaking, rating 7+ systems just aren't really needed that badly, so they aren't very common except for extremely important stuff.

The problem with that Jaid, is that eventually the players will all have 7+ Commlinks, or if they only have a couple, then the hackers will have them.

If they exist, whether they need them or not, VIPs will want them, and have them, even if they don't know how to work them

If they exist, players will get a hold of them, even if they have to plan their own runs to get them.

Or you can just say that rating 6 is the most powerfull system you can have in a portable device and save yourself alot of trouble.

That means the server can be rating 7+, up to godzillion if you care, while your runners will max out at 6 unless they set up a van or something with a computer system in the back of it. Then the hacker has to jack into the computer A'la Matrix style, while the team runs inside the building or whatever.