On Fri, Apr 6, 2012 at 4:17 PM, Olivier Schwander <span dir="ltr"><<a href="mailto:olivier.schwander@chadok.info">olivier.schwander@chadok.info</a>></span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Nice work on the vg2.<br>
<br>
Le 06 Apr 2012 07:58, Shawn Willden a écrit:<br>
<div class="im">> we currently have just over 9 TB available storage space, right around<br>
> 6 TB used, and we're uploading 140 GB per day.<br>
<br>
</div>I'm a bit curious: what are the common use cases for this ? It should be<br>
interesting to have some statistics/examples/success story on the<br>
website.<br></blockquote><div><br></div><div>I think most of us are using it for backup. That was my goal when I started "encouraging" (cajoling?) the early members to establish high uptime, high capacity and wide node distribution requirements -- to create a virtually indestructible Internet-based backup solution.</div>
<div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">> Next goal: 32 nodes highly-reliable nodes.<br>
<br>
</div>What are the advice or requirements to reach your 95% uptime goal ? I<br>
guess the hardware is not really a problem but the connection must<br>
matter a lot. Are you using ADSL, optic fiber, high-end connection in a<br>
datacenter ? I don't have any idea of the availability a home ISP may<br>
provide.</blockquote><div><br></div><div>95% really isn't that high. Over the course of a year, for example, you can have over 18 days of downtime and still reach 95%. That's 36 hours of downtime per month. In practice, most of our nodes exceed the availability requirements by a large margin, even though many of them are on home ADSL or cable modem connections. If your home ISP service were down a day and a half every month, you'd change ISPs. Well, I would, anyway.</div>
<div><br></div><div>The uptime requirement is really more of a question of attitude. We want members of the grid to go into it with the attitude that they have a responsibility to keep their nodes up and running 100% of the time. The stated 95% goal is really just an admission that we understand perfection is impossible. It also provides a baseline we can use for statistical calculations used to estimate failure probabilities and derive reasonable erasure-coding parameters.</div>
<div><br></div></div>-- <br>Shawn<br>