Fault tolerance of colocation switches

Question

Computer networks

Hello dear Khabrovites!
There are a couple of racks in the data center, which are filled with web and database servers, storage systems, etc.
There are 2 Juniper EX3200 switches.
Previously, everything worked simply - everything was plugged into one switch, it also received a cable from the provider and two vlans - for servers looking outside and the internal subnet.
Now I want to make this small network more reliable, so that in the event of one switch failing or the NIC / switch port failing, everything continued to work.
What ideas are there:
from each server, one patch cord enters each switch, bounding is organized on the server side in fail-over mode. The switches are interconnected and each of them receives a cable from the provider's switch.
Cons - no traffic balancing on lvl2.

Please share your best-practice, and also tell me what else needs to be done to make it all take off.

Answer 1

You really need stackable switches, for example, EX3300 and from them there is already LACP to servers and uplinks.
From what there is, you can only fence some kind of STP

Answer 2

We combine switches in VC (via an expansion module, 2x10Ge) in master-master mode (both switches in RE status), we agree with the uplink and make AE. Additional cost for crossover.