<div align="left"><p dir="ltr"><br>
</p>
</div><div align="left"><p dir="ltr">Date: Wed, 26 Oct 2016 09:52:13 -0400<br>
</p>
</div><div align="left"><p dir="ltr">From: Joe Landman <<a href="mailto:landman@scalableinformatics.com">landman@scalableinformatics.com</a>><br>
</p>
</div><div align="left"><p dir="ltr">To: <a href="mailto:beowulf@beowulf.org">beowulf@beowulf.org</a><br>
</p>
</div><div align="left"><p dir="ltr">Subject: Re: [Beowulf] non-stop computing<br>
</p>
</div><div align="left"><p dir="ltr">Message-ID:<br>
</p>
</div><div align="left"><p dir="ltr"> <<a href="mailto:918d4484-89be-0a02-5e86-6041ba31dccf@scalableinformatics.com">918d4484-89be-0a02-5e86-6041ba31dccf@scalableinformatics.com</a>><br>
</p>
</div><div align="left"><p dir="ltr">Content-Type: text/plain; charset=windows-1252; format=flowed<br>
</p>
</div><div align="left"><p dir="ltr"><br>
</p>
</div><div align="left"><p dir="ltr">Licensing might impede this ... Usually does.<br>
</p>
</div><div align="left"><p dir="ltr"><br>
</p>
</div><div align="left"><p dir="ltr"><br>
</p>
</div><div align="left"><p dir="ltr">On 10/26/2016 09:50 AM, Prentice Bisbal wrote:<br>
</p>
</div><div align="left"><p dir="ltr">> There is a amazing beauty in this simplicity.<br>
</p>
</div><div align="left"><p dir="ltr">><br>
</p>
</div><div align="left"><p dir="ltr">> Prentice<br>
</p>
</div><div align="left"><p dir="ltr">><br>
</p>
</div><div align="left"><p dir="ltr">> On 10/25/2016 02:46 PM, Gavin W. Burris wrote:<br>
</p>
</div><div align="left"><p dir="ltr">>> Hi, Michael.<br>
</p>
</div><div align="left"><p dir="ltr">>><br>
</p>
</div><div align="left"><p dir="ltr">>> What if the same job ran on two separate nodes, with IO to local<br>
</p>
</div><div align="left"><p dir="ltr">>> scratch? What are the odds both nodes would fail in that three week<br>
</p>
</div><div align="left"><p dir="ltr">>> period. No special hardware / software required. Simple. Done.<br>
</p>
</div><div align="left"><p dir="ltr">>><br>
</p>
</div><div align="left"><p dir="ltr">>> Cheers<br>
</p>
</div><p dir="ltr"><br>
The Mainframe world went this way. But then had to decide what happens if you get different answers. You need a third node to verify which one is wrong or if it's just a broken application (3 unique answers).<br></p>