How can I prevent a DDOS attack on Amazon EC2?
Solution 1:
A DDOS (or even a DOS), in its essence, is a resource exhaustion. You will never be able to eliminate bottlenecks, as you can only push them farther away.
On AWS, you are lucky because the network component is very strong - it would be very surprising to learn that the upstream link was saturated. However, the CPU, as well as disks I/O, are way easier to flood.
The best course of action would be by starting some monitoring (local such as SAR, remote with Nagios and/or ScoutApp) and some remote logging facilities (Syslog-ng). With such setup, you will be able to identify which resources get saturated (network socket due to Syn flood ; CPU due to bad SQL queries or crawlers ; ram due to …). Don’t forget to have your log partition (if you don’t have remote logging enable) on an EBS volumes (to later study the logs).
If the attack come through the web pages, the access log (or the equivalent) can be very useful.
Solution 2:
You can also further isolate your EC2 instances by putting them behind an Elastic Load Balancer and only accepting traffic from the ELB instance. This puts more of the onus on Amazon to manage DDOS attacks.
I assume that you'll still have SSH open to all, so it's likely you'll still see some rogue traffic coming in there, unless you can lock down that port to some static IPs. You could change the SSHd port to something more obscure (i.e., something other than 22) to further reduce DDOS hits (most bots only check known ports).
I'll also mention fail2ban, which can monitor logs and temporarily modify your ip tables to block specific IPs (for instance, if there have been 6 failed attempts to SSH into your host from a single IP address, it can block that IP for 30 minutes or so). Keep in mind that (as Jordan astutely commented) fail2ban is probably not appropriate for blocking proxied traffic (e.g., that from an ELB) because it will block the IP of the proxy, not necessarily the original remote IP.
I haven't used it, but Apache mod_evasive may also be worth investigating; however, it may have the same weakness as fail2ban when it comes to IP-based blocking.
Solution 3:
If you are using Apache, I suggest using mod_security. Packaged by most vendors, the core rules set does a fantastic job.
Another hardening step is limiting the requests at the webserver level. Nginx., Apache can throttle and limit incoming requests.
Solution 4:
The solution I use for blocking realtime bad activity IP's coming out of AWS and others is this... In my CSF Firewall in the config for LFD Blocklists I use a list found here - http://myip.ms/browse/blacklist/Blacklist_IP_Blacklist_IP_Addresses_Live_Database_Real-time
Download Blacklist for CSF Firewall » http://myip.ms/files/blacklist/csf/latest_blacklist.txt
No more outrageously obnoxious AWS traffic.
Solution 5:
I am biased, as I work for a content delivery network as a security Presales engineer.
However, leveraging a Ddos mitigation solution on a content delivery network ensures that you never run out of resources at the origin. It's similar to putting an F5 load balancer in front of your site, but spread to thousands of locations around the world.
A good cdn will allow you to cloak the origin with a whitelist which you install on the aws firewall. So when attackers perform their reconnaissance on Amazon, your IP address will come up empty as everything will be blocked.
So the Ddos attacks are blocked when the traffic hits a node as close as possible to the attacker. This ensures you mitigate Ddos attacks as far away from the asset you are trying to protect.
A good cdn can also perform health checks and failover traffic to other locations eg another ego on in ass, Azure, rack space, soft layer, a physical dc etc. It should also have a WAF to ensure you can block application layer exhaustion attacks like RUDY, slowpost, slowloris as well as sqli, xss, rfi, lfi etc.
By default the cdn also blocks network layer attacks like teardrop, icmp attacks, synfloods etc. A cdn is able to mitigate Ddos attacks because trey have vast amounts of capacity to accept the requests, filter out bad traffic and pass on the good traffic. So amplifying attacks like ntp, DNS, ssdp, chargen and snmp volumetric attacks can be blocked.
The largest attack I have seen to date has been 321gbps in July 2014. During this attack there was also a DNS protocol attack at 20gbps. So you will need to ensure you DNS infrastructure is also resilient to withstand a huge number of requests.
From the description you provided, it looks like you were subject to an exhaustion attack, where thr attacker opened up lots of threads such that all the threads were consumed up on the web server, app server or firewall. It's similar to something like a slowpost, slowloris or RUDY.
To block against application layer exhaustion attacks you will need to get a web application firewall (WAF). A typical network firewall (including amazons firewall and next generation firewalls) won't be able to block it. Sent work firewalls these days can only block about 30% of all the attacks these days (Nov 2014).