First, you'll need to determine what it means by "ISP can temporary increase our speed up to 70mbps." Is the 70 Mbps always guaranteed when you need it? If it is does, whether going beyond 40 Mbps, subjects the "excess" to higher drop probability and/or lower transit priority across ISP infrastructure? Further, you may want to consider if, when you exceed 40 Mbps, whether there's a surcharge, especially a "high rate" surcharge.
Depending on the answers to those questions, you may need to limit your usage to your guaranteed 40 Mbps to insure you'll always have your 30 Mbps and at an expected cost.
Whether you set your overall limit to 40 Mbps, or 70 Mbps, the QoS to accomplish what you want is much like what @Georg Pauwen already posted, but I would recommend a few changes and notes to Georg's recommendation, such as:
access-list 101 permit ip 192.168.1.0 0.0.0.255 any
!
class-map match-any CUSTOMER_30_MBPS
match access-group 101
!
policy-map CHILD_SHAPER
class CUSTOMER_30_MBPS
priority 30720
bandwidth 30000 !I recall (?) bandwidth is in Kbps
fair-queue
class class-default
bandwidth 10000 !or 40 Mbps, i.e. all classes, together, should add up to total available, usable, bandwidth
fair-queue
!
policy-map PARENT_SHAPER
class class-default
shape average 70000000 !(I recall [?] shapers, or policers, use bps) or 40 Mbps, BTW I suspect some shapers don't account for L2 overhead, if not reduce rate by about 15%
service-policy CHILD_SHAPER
