WP Breaking Down Mirai: An IoT DDoS Botnet Analysis

Breaking Down Mirai: An IoT DDoS Botnet Analysis

Breaking Down Mirai: An IoT DDoS Botnet Analysis

By now many of you have heard that on September 20, 2016, the website of renowned security journalist Brian Krebs was hit with one of the largest distributed denial of service attacks (DDoS) to date.

The magnitude of that attack, the star status of its target within the InfoSec community and the heaps of drama that followed made this one of the most high-profile DDoS stories of the year.

On September 30, the story saw another development when a HackForum user by the name of ‘Anna-senpai’ leaked the source code for Mirai—the botnet malware behind the attacks. It was speculated that in doing so the perpetrator was trying to hide his tracks, rightfully concerned about the repercussions of taking a swing at Brian.

Since the source code was published, the Imperva Incapsula security team has been digging deep to see what surprises Mirai may hold. In this post we’ll share:

  • Our own encounters with Mirai botnets
  • The results of our investigation of Mirai’s source code


New Mirai scanner released: We developed a scanner that can check whether one or more devices on your network is infected by or vulnerable to Mirai. You can find the beta of the Mirai Scanner here.

If you missed out “Deep Dive into the Mirai Botnet” hosted by Ben Herzberg check out our video recording of the event.

Close Encounters of the Third Kind

A thorough review of Mirai’s source code allowed us to create a strong signature with which we could identify Mirai’s activity on our network. We then turned to our logs and examined recent assaults to see if any of them carried Mirai’s fingerprints.

Sure enough, we found the Mirai botnet was responsible for a slew of GRE floods that were mitigated by our service on August 17. Using a hit-and-run tactic, the attack peaked at 280 Gbps and 130 Mpps, both indicating a very powerful botnet.

Mirai-powered GRE floods

Figure 1: Mitigating a slew of Mirai-powered GRE floods, peaking at 280 Gbps/130 Mpps

Investigation of the attack uncovered 49,657 unique IPs which hosted Mirai-infected devices. As previously reported, these were mostly CCTV cameras—a popular choice of DDoS botnet herders. Other victimized devices included DVRs and routers.

Overall, IP addresses of Mirai-infected devices were spotted in 164 countries. As evidenced by the map below, the botnet IPs are highly dispersed, appearing even in such remote locations as Montenegro, Tajikistan and Somalia.

Geo-locations of all Mirai-infected devices

Figure 2: Geo-locations of all Mirai-infected devices uncovered so far
Country % of Mirai botnet IPs
Vietnam 12.8%
Brazil 11.8%
United States 10.9%
China 8.8%
Mexico 8.4%
South Korea 6.2%
Taiwan 4.9%
Russia 4.0%
Romania 2.3%
Colombia 1.5%
Figure 3: Top countries of origin of Mirai DDoS attacks

Interestingly, since the source code was made public, we’ve also seen a few new Mirai-powered assaults. This time they took the form of low-volume application layer HTTP floods, one of which was even directed against our domain (www.incapsula.com).

Characterized by relative low requests per second (RPS) counts and small numbers of source IPs, these looked like the experimental first steps of new Mirai users who were testing the water after the malware became widely available. Likely, these are signs of things to come and we expect to deal with Mirai-powered attacks in the near future.

 Mirai botnet HTTP flood

Figure 4: Mirai botnet launching a short-lived HTTP flood against incapsula.com

Source Code Analysis

Mirai is a piece of malware that infects IoT devices and is used as a launch platform for DDoS attacks. Mirai’s C&C (command and control) code is coded in Go, while its bots are coded in C.

Like most malware in this category, Mirai is built for two core purposes:

  • Locate and compromise IoT devices to further grow the botnet.
  • Launch DDoS attacks based on instructions received from a remote C&C.

To fulfill its recruitment function, Mirai performs wide-ranging scans of IP addresses. The purpose of these scans is to locate under-secured IoT devices that could be remotely accessed via easily guessable login credentials—usually factory default usernames and passwords (e.g., admin/admin).

Mirai uses a brute force technique for guessing passwords a.k.a. dictionary attacks based on the following list:

root     xc3511
root     vizxv
root     admin
admin    admin
root     888888
root     xmhdipc
root     default
root     juantech
root     123456
root     54321
support  support
root     (none)
admin    password
root     root
root     12345
user     user
admin    (none)
root     pass
admin    admin1234
root     1111
admin    smcadmin
admin    1111
root     666666
root     password
root     1234
root     klv123
Administrator admin
service  service
supervisor supervisor
guest    guest
guest    12345
guest    12345
admin1   password
administrator 1234
666666   666666
888888   888888
ubnt     ubnt
root     klv1234
root     Zte521
root     hi3518
root     jvbzd
root     anko
root     zlxx.
root     7ujMko0vizxv
root     7ujMko0admin
root     system
root     ikwb
root     dreambox
root     user
root     realtek
root     00000000
admin    1111111
admin    1234
admin    12345
admin    54321
admin    123456
admin    7ujMko0admin
admin    1234
admin    pass
admin    meinsm
tech     tech
mother   f**er [censored]

Mirai’s attack function enables it to launch HTTP floods and various network (OSI layer 3-4) DDoS attacks. When attacking HTTP floods, Mirai bots hide behind the following default user-agents:

Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/51.0.2704.103 Safari/537.36
Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/52.0.2743.116 Safari/537.36
Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/51.0.2704.103 Safari/537.36
Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/52.0.2743.116 Safari/537.36
Mozilla/5.0 (Macintosh; Intel Mac OS X 10_11_6) AppleWebKit/601.7.7 (KHTML, like Gecko) Version/9.1.2 Safari/601.7.7

For network layer assaults, Mirai is capable of launching GRE IP and GRE ETH floods, as well as SYN and ACK floods, STOMP (Simple Text Oriented Message Protocol) floods, DNS floods and UDP flood attacks.

Mira also seems to possess some bypass capabilities, which allow it to circumvent security solutions:

#define TABLE_ATK_DOSARREST                     45  // "server: dosarrest"
#define TABLE_ATK_CLOUDFLARE_NGINX      46  // "server: cloudflare-nginx"
if (util_stristr(generic_memes, ret, table_retrieve_val(TABLE_ATK_CLOUDFLARE_NGINX, NULL)) != -1)
                        conn->protection_type = HTTP_PROT_CLOUDFLARE;
if (util_stristr(generic_memes, ret, table_retrieve_val(TABLE_ATK_DOSARREST, NULL)) != -1)
                        conn->protection_type = HTTP_PROT_DOSARREST;

While this may seem like a standard source code, Mirai also has a few quirks that we found especially intriguing…

Mirai’s “Don’t Mess With” List

One of the most interesting things revealed by the code was a hardcoded list of IPs Mirai bots are programmed to avoid when performing their IP scans.

This list, which you can find below, includes the US Postal Service, the Department of Defense, the Internet Assigned Numbers Authority (IANA) and IP ranges belonging to Hewlett-Packard and General Electric.               - Loopback                 - Invalid address space                 - General Electric (GE)                - Hewlett-Packard (HP)                - US Postal Service                - Internal network            - Internal network             - Internal network             - IANA NAT reserved            - IANA NAT reserved             - IANA Special use
224.*.*.*+                - Multicast                 - Department of Defense                - Department of Defense                - Department of Defense                - Department of Defense                - Department of Defense                - Department of Defense                - Department of Defense                - Department of Defense                - Department of Defense               - Department of Defense

This list is interesting, as it offers a glimpse into the psyche of the code’s authors. On the one hand, it exposes concerns of drawing attention to their activities. A concern we find ironic, considering that this malware was eventually used in one of the most high-profile attacks to date.

On the other hand, the content list is fairly naïve—the sort of thing you would expect from someone who learned about cyber security from the popular media (or maybe from this Wiki page), not a professional cyber criminal.

Together these paint a picture of a skilled, yet not particularly experienced, coder who might be a bit over his head. That is unless some IP ranges were cleared off the code before it was released.

A Territorial Predator

Another interesting thing about Mirai is its “territorial” nature. The malware holds several killer scripts meant to eradicate other worms and Trojans, as well as prohibiting remote connection attempts of the hijacked device.

For example, the following scripts close all processes that use SSH, Telnet and HTTP ports:

killer_kill_by_port(htons(23))  // Kill telnet service
killer_kill_by_port(htons(22))  // Kill SSH service
killer_kill_by_port(htons(80))  // Kill HTTP service

These locate/eradicate other botnet processes from memory, a technique known as memory scraping:

#DEFINE TABLE_MEM_UPX             // \X58\X4D\X4E\X4E\X43\X50\X46\X22

And this function searches and destroys the Anime malware—a “competing” piece of software, which is also used to compromise IoT devices:

searching for .anime process
                // If path contains ".anime" kill.
                if (util_stristr(realpath, rp_len - 1, table_retrieve_val(TABLE_KILLER_ANIME, NULL)) != -1)
                    kill(pid, 9);

The purpose of this aggressive behavior is to:

  • Help Mirai maximize the attack potential of the botnet devices.
  • Prevent similar removal attempts from other malware.

These offensive and defensive measures shine a light on the turf wars being waged by botnet herders—a step away from the multi-tenant botnets we previously encountered in our research. So much for honor among thieves.

From Russia with Love?

Lastly, it’s worth noting that Mirai code holds traces of Russian-language strings despite its English C&C interface. Here, for instance, Russian is used to describe the “username” and “password” login fields:

// Get username
this.conn.SetDeadline(time.Now().Add(60 * time.Second))
this.conn.Write([]byte("\033[34;1mпользователь\033[33;3m: \033[0m"))
// Get password
this.conn.SetDeadline(time.Now().Add(60 * time.Second))
this.conn.Write([]byte("\033[34;1mпароль\033[33;3m: \033[0m"))

This opens the door for speculation about the code’s origin, serving as a clue that Mirai was developed by Russian hackers or—at least—a group of hackers, some of whom were of Russian origin.

Other bits of code, which contain Rick Rolls’ jokes next to Russian strings saying “я люблю куриные наггетсы” which translates to “I love chicken nuggets” provide yet more evidence of the Russian heritage of the code authors, as well as their age demographic.

What Can You Do to Prevent IoT Botnet from Spreading

While DDoS attacks from Mirai botnets can be mitigated, there’s no way to avoid being targeted. However, as a device owner, there are things you can do to make the digital space safer for your fellow Internet citizens:

  • Stop using default/generic passwords.
  • Disable all remote (WAN) access to your devices. To verify that your device is not open to remote access, you can use this tool to scan the following ports: SSH (22), Telnet (23) and HTTP/HTTPS (80/443).

With over a quarter billion CCTV cameras around the world alone, as well as the continued growth of other IoT devices, basic security practices like these should become the new norm. Make no mistake; Mirai is neither the first nor the last malware to take advantage of lackluster security practices.