Boost plugin disk perfomance issue.

[Akven]

Good evening.
We are using Cacti 0.8.8b and Boost plugin 5.1
And we're encountering some performance hit to polling when Boost updates RRAs.

SYSTEM STATS: Time:2.2848 Method:spine Processes:2 Threads:20 Hosts:951 HostsPerProcess:476 DataSources:1606 RRDsProcessed:0

Data Sources:
Script/Command: 3
SNMP: 956
SNMP Query: 1096
Script Query: 2
Script Query - Script Server: 1
Total: 2058

Linux cacti 3.13.0-37-generic #64-Ubuntu SMP Mon Sep 22 21:28:38 UTC 2014 x86_64

Normaly Cacti works fine. But when Boost period expires - during the RRD Update time appears gaps on graphs, because Poller begins to exceed the interval.

12/17/2014 01:29:57 PM - SYSTEM BOOST STATS: Time:2.4300 RRDUpdates:7161
12/17/2014 01:31:02 PM - POLLER: Poller[0] Maximum runtime of 58 seconds exceeded. Exiting.
12/17/2014 01:31:23 PM - SPINE: Poller[0] ERROR: Spine Timed Out While Processing Hosts Internal

We tried several different Boost options. Tried 30 minutes update or 4 hours or even limited humber of maximum records, so boost updates every 10 minutes. No effect. Gaps still appear.
Maximum Data Source Items Per Pass currently on 2000 data source items.


Server does not have good storage hardware. It is 2 Cores 2Gb RAM VMWare Sphere on some virtual server. I do not have access to hardware itself. But it seems it cannot handle the I/O and affecting polling that way. I see a lot of disk usage and latencies on VMSphere monitoring during updates.
Is there any way to set the priority to Boost updates or even limit the rate at which the Boost updating files. Or limit the number of files it updates simultaneously.
Maybe if I can lower Maximum Data Source Items Per Pass to 500 items it could help?
I do not mind if Boost would collect data for 30 minutes and then slowly update files for 20 minutes if it would eliminate gaps. But there is no such functionality in plugin.