1 Aug 2011

Review: PingCraft LagKiller (Part 1)

Warcraft can be a challenging game at the best of times. Whether it’s fighting raid bosses, battling it out in arenas or just trying to cram daily quests into our hectic schedule we all want to be playing at our best.

There’s various things as gamers that we can do to improve our game time. One is making sure we’re using a computer with enough grunt to handle our favourite MMOs. The next is to select keyboards, mice and other peripherals that support our individual styles of play.

The final piece is our internet connection to the game servers. This is the one thing we sometimes struggle with. You can spend ages trying to pick a good ISP or even move house to get better speeds. Sometimes though there’s pretty much nothing you can do.

This is where latency reduction and tunnelling services like PingCraft’s LagKiller come in – aiming to reduce your in-game latency for a small monthly fee. But does it actually work or is it just some snake-oil scam? Time to put it to the test.

Services like LagKiller aim to reduce network latency through a combination of tunnelling and proxies, redirecting your Warcraft traffic through their server network instead of relying on your ISPs peering arrangements. These types of service aren’t against Blizzard’s Terms of Use (WoW Insider’s Lawbringer series did an article featuring them recently), but they can trigger an account lock through false positives.

Although LagKiller is designed to work with any MMO that uses TCP, I decided to stick to using World of Warcraft for testing purposes. This is partly so I could build up a healthy stack of data, but also I could use the Addon framework in order to measure ping times as reported in-game. I used a modified version of shPerformance in order to measure latency and bandwidth every 30 seconds (as often as Warcraft updates these values) and output the results to a chat channel. I then used WoWScribe to record these values.

World of Warcraft reports two ping values. For the purpose of this test I’ve logged and compared both types to provide more in-depth analysis.

  • Home ping is your connection to the realm server (Earthen Ring EU in this case). Chat data, guild chat and info, auction house data and so on are all sent using this connection. Unless you’re doing an auction house scan this connection is usually fairly lightweight.
  • World ping is your connection to the environment server. Information on combat, nearby player data (specs, gear and enchants) and NPCs (vendors, quest givers, hostile mobs etc) all use this connection. It’s also used for casting and professions – basically everything that doesn’t involve the realm server.

Starting the Test

Testing was done using a single PC running Windows 7 Ultimate Service Pack 1. You can find more information about the test PC used here. Real-world conditions were used – I performed daily quests in the Molten Front, ran heroic dungeons until my eyes bled and sat around in Stormwind waiting for an auction house scan to finish or the LFG queue to pop. I always connected to the same realm (Earthen Ring EU) as not all realms are in the same datacentre. Wowpedia has lists of both EU realms and US realms by datacentre.

I should also declare some factors that may influence test results. As I was only using a single PC, testing was not simultaneous and did take place at different times  for when the service was enabled and disabled. All sample collection was performed midweek at the same time of day but on different days, so network variances could creep in. This also means that sample sizes are not identical. If you’re interested in testing it out for yourself, PingCraft do also provide a limited trial of their service, allowing you to check it against your own setup before taking the plunge.

For the occasions where I was using LagKiller, I allowed it to automatically select which servers to use and left it to run automatically. When in use it also provides its own bandwidth monitor and some details on which servers it has connected to.

Why go to all this effort to measure any change? I’m one of the fortunate few who has a good quality service through a great ISP. I also chose where I live based on the distance from the telephone exchange, so I get a superfast connection as standard. As I couldn’t just rely on perceiving any difference in play I needed to have something to measure ping more accurately.

The Results

I’ve broken down the results into three categories were the majority of testing was carried out. I’ve presented the results using a percentile Cumulative Histogram in order to allow comparison between when the service is in use and when I’ve disabled it. I’ve used this method rather than comparing averages as ping times can vary wildly during play.

The first set of results (and the easiest to process) are from me being sat around in Stormwind City. You can see that while I have a pretty good connection my ping times (in blue) can vary wildly, especially to the World server due to the sheer number of other people there doing their own thing.

By contrast, using LagKiller (in green) had a major effect. While my lowest measured ping is still the same (as I said, I have a great connection normally), the overall variance of pings is hugely reduced and the connection is tightened up overall.

Sitting around in Stormwind is all well and good, but how about some actual combat? How about the Molten Front daily quest hub? For these tests I measured latency while carrying out the various daily quests inside the Molten Front only (excluding the time spent outside in Mount Hyjal).

Again, it’s easy to see the difference that LagKiller has made to the connection. Although the lowest ping time is the same the variance is much tighter with the service enabled.

To round of this batch of tests I recorded some test data as I went through the process of completing LFG heroics for those highly desirable valour points. It’s this test that produced the most interesting set of results so far.

The first thing you’ll notice is that all four curves are much tighter as my good quality connection handles the restricted amount of players and carefully controlled combat levels easily. You’ll also notice that LagKiller continues to provide the best overall result and showing the lowest variance in ping times.

The curious artifact is the ‘scoop’ in the PingCraft Ping(world) trace from about 80%. I’m not sure what’s causing this and would need to investigate it further. Current theories suggest that it’s due to the way LFG groups are assembled. As we only have one LFG pool in the EU, it’s not clear which datacentre is selected to host that LFG instance. As such, even though mu connection to the Frankfurt datacentre is used for my realm, the LFG session may be held in Paris. As Paris is closer than Frankfurt to me, it could possibly generate this kind of result.

The observant among you will have probably noticed that there’s two areas I’ve not tested. The first one is raiding, and I’ve not presented plots of results from Firelands raids purely due to sample size. I figured that the best numbers would come from real world raids, not from kiting scorpions or turtles round in circles in the hope of epics. I’ll hopefully be reporting these in a Part 2 later in the month once I’ve gathered more data.

The second one is arenas, and this is purely down to me not being that great a PvPer and not being in a team. If I can amend the team bit (there’s no hope for me skill wise) then I’ll also try and capture some stats for them.

I’m also going to try and capture some data from playing transatlantic and using a US WoW account. While the data collected here will probably be fairly basic, it should help to amplify any differences for those playing on servers overseas.

In Conclusion

To sum up, I’m actually surprised by the results seen so far. I came into this being sceptical of the service and assuming it was some kind of scam. The results speak for themselves though – there’s clear evidence that the service does work and could clearly be a benefit to those suffering with poor connections. If you are suffering from lag issues then I suggest giving the trial a go and see if it fixes the issue for you.

In terms of cost, PingCraft offers LagKiller on a monthly subscription basis. The basic service comes in at $5.99 a month, while the advanced service (tested here) is $9.99 a month. Which one you’d want depends on your needs, with the advanced one offering more customisation options should you need them.

Look forward to Part Two later in the month, where I’ll be answering questions about raiding and cross-continent performance. In the meantime, feel free to hit me with any questions or thoughts in the comments section below.

Note: Mana Obscura was provided with free use of the service for the purpose of preparing this review.

Like this? Try these other related posts:

Tags: , , , , ,

10 Responses to Review: PingCraft LagKiller (Part 1)

  1. Ln says:

    I’m curious as to how these services work, so I’ll run through my thought process in the hope that someone can help me out!

    Logically, you’re not actually reducing your real-world latency. What I mean is this:

    When this Lagkiller is installed it sets up a proxy for your WoW client to a server which has a super fast dedicated network pipe at some datacentre with good connectivity back to Blizzard’s own DC. Essentially, you’re sending your traffic to this proxy, which is then forwarding the traffic to Blizzard on your behalf. Blizzard believe that the data it’s being sent is coming from the proxy server, not your own PC. Blizzard then reply to the proxy server, and finally the proxy server forwards Blizzards replies back to you. It just acts as a halfway house, so to speak.

    So, you look at the latency report in-game, and it’s telling you that your latency is, for example, 100ms less than normal. Great! However, is this not just reporting what the latency is to the proxy server (which is then reporting that same value back to you)? Of course it’s going to be lower – the proxy server has much better connectivity than your home does, and depending on where it’s based, it’s own datacentre might well have a direct peering agreement with Blizzard’s datacentre. So, that latency report you’re seeing doesn’t actually include the time it takes for the proxy server to send the traffic back to your own machine and vice versa – that latency still exists, right?

    I’m probably missing something here, but the way I see it is that you’re not actually reducing your latency at all, you’re just making your realm BELIEVE that your latency is lower. And, you know what, that might be all it takes to make your experience better – perhaps Blizzard’s realm places more “trust” in your key presses or actions if it thinks that your latency is lower, and maybe this helps when they calculate what’s happening & when on the server? I don’t know. All I do know is that I’m confused as to how these services can work like they say they do!

    • Gazimoff says:

      It’s a good analogy, but not quite there.

      In normal situations my connection to the server is routed through a series of notes or hops, depending on the peering arrangement they have with internet backbone providers. This is why a server closer to you has lower ping – it has fewer hops to go through before it reaches it’s destination.

      Tunnelling works differently to just plain proxying – your packets enter one end of the tunnel near where you are, and emerge from the tunnel close to the Blizzard datacentre. It routes and prioritises the traffic differently, meaning that it’s likely to get there quicker.

      It’s also why I used the actual in-game latency meter rather than anything else, as it’s measuring the connection to the servers directly and locally – it’s not measured by the server and then sent back. It’s the closes I can get to measuring the difference without resorting to a network analyser (a very expensive piece of kit)

  2. Ln says:

    Ah! Ok, that makes much more sense! How does it create this tunnel, though? Is it simply a VPN service that is invoked when you run the service?

    Let’s take the realm servers in Frankfurt. I’m with Virgin Media, who peer with TeliaSonera (Blizzard’s IP space provider) at LINX Brocade and Extreme, and also DE-CIX in Frankfurt (where, presumably, they also peer with Blizzard’s DC host, Interxion, I’m guessing?). That’s already a pretty direct route to Blizzard (VM’s network into Telia’s network), especially when you consider that VM have a 40Gb connection at DE-CIX and 30Gb at both LINXs.

    I guess it gets around VM’s traffic shaping which could well improve latency significantly (especially at the moment, with all the problems VM are experiencing on a regular basis), but I don’t see how this tunnel could have a more direct route to Blizzard than a large ISP with a direct peering agreement does.

    Sorry for the questions! I’m just trying to get my head around it, haha. Thanks very much for your reply 🙂

  3. WTFastPWNer says:

    LagKiller helps improve your game traffic 2 ways; 1. It gives your game traffic a higher priority with your Internet provider. 2. It gives your game traffic a faster and more stable route to the game server.

    It really does work and it is not smoke and mirrors or some trickery. Give the trial a go and you can see for yourself!

  4. datgrl says:

    Sorry, I’m a dummy 🙁 What are those graphs showing? Frames per second? Ping success rate? Ping response in milliseconds? I’m not sure that a ping test is adequate for a service like WoW that uses a different protocol. The only real way to be sure is to measure an actual packet that contains game information from start to finish. Something that a higher layer performance tool would do.

    • Gazimoff says:

      Good question!

      What I did was measure the in game Latency that WoW reports itself – if you hover over the computer icon in the minimenu in game you can see the latency reported. This isn’t a plain ping test – it’s based off the actual server/response for Blizzard’s own protocols.

      Those latency readings get updated every 30s by the game client, and are exposed to addons via the LUA API. I just used an addon to poll for this measurement every 30s and log it.

      From there I plotted a percentage based cumulative histogram of the latency recorded every 30s. The Wikipedia article I linked to explains what cumulative histogram means far better than I could, but essentially the more vertical the line the smaller the distribution and the more consistent the connection to the game servers.

      I’ve probably taken a bit of a liberty by transposing latency for ping on the chart legends, and I apologise for that. Rest assured though that this is based off in-game latency as reported by the game itself. You can find more information about the API call used here: http://www.wowpedia.org/API_GetNetStats

  5. Abhorsen says:

    Could one potentially use this program to play the game on a network where it is usually blocked? Or have I misunderstood the way it works 😮

  6. Gandalf says:

    hi i’m using a wireless connection on swtor at the moment. I’m located far away from my router so wireless is the only option for me. I also use BT openzone to play past 11 PM. I have these lag spikes when i enter combat putting my 40-50 ping up to 70-110 which can kill me in pvp fights and lower my DPS in raids and flash points. I used the free trial but it does’t realy last long enough for me to test it properly as i cannot do Flash points due to the DC after 20 mins. I was wondering if this software works with wireless connections and if it matters if they are router or wifi access points. And does the premium service increase speeds more in comparison to the premium service?

  7. Gandalf says:

    sorry i was meant to wright premium in comparison to standard*