Why exactly do you need more than 512MB Burst? You should never rely on burstable RAM for your site's usage. It should only be there as a premium to be able to use in peaks.
Because most vps companies equate burst to privvmpages under the vz architecture, and most applications allocate far more memory then they use. 512MB burst is not very high, and will cause allocation problems when running many processes. If you are running xen this is not the case so disregard.
real example i just captured for you:
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
named 7347 0.1 0.2 39540 3084 ? Ssl 22:07 0:00 /usr/sbin/named -u named
F S UID PID PPID C PRI NI ADDR SZ WCHAN TTY TIME CMD
1 S 25 7347 1 0 78 0 - 9885 rt_sig ? 00:00:00 named
RSS = ~40MB, SZ=~10MB
Do you want to be accounted/limited on 40MB or 10MB for named's memory 'usage'? If your burst is not substantially higher then the guarantee, in many cases you won't even be able to use the guarantee. I have seen some vps companies ubc configs 'cheat' (usually they do not know much about how linux accounts for memory) customers with this tactic, as they are guaranteeing memory that under real production use, you will not be able to use your guarantee. Imagine if your vps was running all copies of bind, your burst would need to be 4X your guar just to use your guar. if you are running software that malloc's exactly the amount of space needed for storage, you can have a 'guarantee=burst limit' just fine.
hope this clears up how vz and linux accounts memory 'usage' (there is no perfect accounting formula). it is a heavily debated concept on lkml, linux-containers, and ckrm.
If you look at a java runtime, the allocation is insane, like 400MB wheras the 'usage' is like 60MB.