Most of us detest the idea of having a huge Cisco switch that sounds like an airplane, costs almost as much as an airplane and weighs like one too (well maybe not quite, but you get the idea). However, there are some more suitable alternatives out there for the small home network.

I stumbled upon these nice switches from Mikrotik. The RB260GS for instance features pretty much everything you would ever need. 5 Gigabit Ethernet ports + 1 fixed SFP Gigabit cage all in a tiny plastic chassis the size of a small WLAN router. It runs SwOS which lets you easily set up port forwarding and port mirroring. It supports VLANs, jumbo frames, and lets you set up some nice bandwidth limitation stuff too. All this for $39.95.

Ive recently installed mine and the provided web interface is smooth and extremely simple. No cumbersome Cisco-style command line kung-fu needed here. I had everything set the way I wanted within a minute or two.

This enables me to have a span port which is fed into my soon-to-be server which will let my SecurityOnion sensors monitor all traffic coming and going on my network.

I want to set up a server at home. The server is going to run ESXi and run a few VMs. SecurityOnion for monitoring the home network (meaning we need two ethernet cards in there), a Linux server that will serve the role of NAS, a service server (Arch Linux) that will offer things such as NTP, and some other minor stuff. Finally, a Cuckoobox install. I am sure I will think of more uses in the future, but I have a general idea of what I want at least.

So, summarized, the goal here is as follows:

• Quiet
• High performance
• Small form factor
• Effective
• Two network cards that work well with ESXi

Lets get started.

First, the case. I have decided to opt for the Fractal Design Node 304. It offers excellent air flow, a modular interior with support for up to 6 (!) hard drives in a mini-itx format. Its not too expensive either. Its a bit larger than the smaller cases in its class, but considering what I am using it for, its not unreasonable by any means.

By going for the Node 304, I need to take in consideration that the PSU needs to be 160mm or shorter in depth. Even less for a modular one, which I would ideally want. After some decent research, I discovered that the SilverStone SFX ST45SF-G is a perfect match. Its really small but offers 450 watt, which is more than sufficient for my setup considering I wont need a dedicated GPU. Its 100% modular and packs it all in 125 mm (W) × 63.5 mm(H) × 100 mm(D). I dont think I will find a more suitable power supply than this.

The motherboard is an important one. I looked at quite a few. I was looking for one that has two integrated ethernet cards. ASUS, Gigabit and others did have a few good models but they almost exclusively make use of Realtek cards, with is really something you want to avoid if you’re running ESX. Avoid in general is my recommendation really. I then stumbled upon a really nice Intel board. The Intel DQ77KB! This blog post really won me over. Two intel gigabit ethernet cards in there too along with VT-d support, Ivy Bridge Q77 chipset, and decent connectivity with multiple USB 3.0 ports, HDMI, and pretty much everything else I would ever need. However, this board poses two important restrictions on my setup. For RAM I will be limited to laptop-style DDR-3 SO-DIMM 1600/1333MHz RAM. This, due to the fact that there’s only 2x RAM slots on the board, limits me to a maximum amount of 16GB RAM. A tough one to swallow for an ESX-server. The real question is if the Intel board would technically support more than 16GB once 16GB SO-DIMM chips become available or not. I eventually concluded that 16GB of RAM will have to do. It will still allow me to run 8 VM’s with 2GB of allocated RAM, which should have me covered for now. Its simply one of the sacrifices necessary to get the small form factor. The second major limitation is the CPU. Intel states that the DQ77KB supports CPU’s with up to 65W TDP. To be honest, its not really a significant restriction for me as the i7-3770S is officially supported by the board and provides more than ample performance. I guess ideally, a really beefy Xeon CPU would be ideal, but I think my 5-6 VMs should be able to cope pretty well. They wont be running particularly intensively, at least not at the same time.

So, for RAM I ended up with a standard kit Crucial. 16GB, 1600MHz, CL11. An the Intel i7-3770S will slip into the CPU socket. It has slightly lower clock speed than the 3770, but it gives me 8 cores which is what really matters with ESX. Alternatively, I could opt for the 3770T which has an even lower TDP at 45W, but availability is still an issue here and performance should be significantly lower than the 3770S, but that might be a trade-off I am willing to make.

For harddrives I intend on having an SSD host the OS and WD Red-series drives take care of file storage. Initially, mostly to save money I will go with one of the 3TB drives. I dont need more than 3TB for now and I can easily upgrade later. The Red series is designed to run 24/7 and is probably the most reliable realistic option in my case.

The CPU cooler needs to be silent and low profile. I could opt in for water cooling, but for now I think I will settle for the Noctua NH-L9i. It seems effective enough and with a specified 14-23dbA, its quiet too. It does include a Low-Noise adapter which can be used on CPUs with TDP less than 65W. I have to do some more research before determining whether I can actually use it with the 3770S, but if that turns out to be the case, then it should be near-silent.

The parts are soon to be ordered and once they arrive, I will do a post on how things turn out. Meanwhile, if anyone have any feedback or suggestions, please leave a comment here to help me out. Any modifications I do to the setup will be reflected in updates to this post.

YubiNotes is an Open Source secure note app for Android. Notes are not left unencrypted on the device for anyone to grab and are not pushed out into the big cloud for the whole world to see. Moreover, it supports the YubiKey NEO which will let you encrypt and decrypt your notes simply by tapping the YubiKey on your NFC capable Android device. Dont worry if you’ve yet to obtain a YubiKey though, the app can be used perfectly fine with a regular password too.

The app is coming along nicely and will hopefully be released upon the Google Play Store later this week. However, it has already won a nice award in the Yubi March 2013 competition. If you have any questions, feel free to shoot them off in the direction of the comments.

If you’ve worked with Android app development, you’ve most likely come across intents and ContentProviders. These are actually just wrappers around the Binder framework.

So with Binder, you typically have a server and client process and want them to communicate. We need a mechanism between the two to make that work. You can use Binder to facilitate that communication. Binder is used so extensively in the Android framework that its amazing how little documentation there is on it.

To help you along, Ive written an example application. Its very simple and all it really does is write custom log messages. However, it consists of a client, a server and a library project that works together using Binder. Take a look at the code and see if you cant learn a thing or two. Have fun!

UPDATE: 1 copy given out, so there’s two more to be handed out.

fastboot flash boot boot.img

.. or take it for a test run:

fastboot boot boot.img.

I committed the code to CM, so if you want to take a look, check out the gerrit entry.

At home I have a very simple setup. I have a Netgear WDR3700v1 WLAN router. To that I have my media center system and main desktop system connected through ethernet, while everything else goes over WiFi. The WDR3700 is pretty decent and allows neat things such as guest networks, network isolation, etc. However, as with nearly every consumer router out there, it does not support more advanced features such as setting up a SPAN port.

Now in this first part of my little project, I am going to flash OpenWRT, an open source Linux based firmware for embedded devices. It gives you full control and allows you to do an endless amount of things that your standard router firmware will never allow. Package management, packet scheduling, IDS/IPS, AQM, statefull firewall, Dynamic DNS, port knocking, load balancing, and the list goes on and on. My personal needs are fairly simple. I want to set up full network monitoring using Snort and for that I simply need to dedicate one of the 4 ethernet ports on the router for SPAN, allowing all traffic going through the network to dump out. At the time of writing, I havent quite decided if I want to turn my IDS into an IPS, in which would make things more complicated, but I am fairly certain I want to keep things simple and stick with IDS for now.

At any rate, the first order of business, and the main purpose of this part of the guide, is to install OpenWRT on my WDR3700. OpenWRT supports an impressive array of devices. To see if your router is supported, simply check the hardware table. The WDR3700 is fully supported, and instead of building the source, I can simply grab a precompiled image. If you’re coming from a stock firmware, then the factory images are the ones you want. If you’re coming from an earlier version of OpenWRT, then the sysupgrade images is the right pick. Only real difference is in the headers anyway. The WDR3700 uses an Atheros AR7161 card and the latest version of OpenWRT is available from the download site.

So, onto flashing the firmware. Its actually dead simple in most cases. However, the process differs from router to router, so you should definitely check out the OpenWRT wiki pages for detailed instructions for your specific router. In most cases, all you have to do is simply login to the router administration web interface, upload the firmware file as you would do with a regular stock firmware upgrade, and flash away. However, make sure you have your ethernet cables at the ready, because things such as WLAN might not work out of the box without the required kernel modules (which is where the aforementioned package manager comes in handy).

Uploading the image worked flawlessly, but after that, things were pretty quiet. Quiet in the sense that right now you have basically reduced your fully working WLAN router to a simple computer running pretty stripped down Linux distribution. So first order of business is to follow the getting started guide and set up necessary configurations.

Now, to get WLAN up and running, I needed some kernel driver modules. These were easily grabbed using the packet manager. I SSH’ed into the box (after telnetting in and setting a password, which is necessary for getting SSH working), and grabbed the necessary packages.

# opkg update
# opkg install kmod-ath9k wpad-mini

After restarting the router, wireless was essentially working. Though configuration remains.

Firstly though, I needed to get a layout of the switch ports and figure out which ports mapped to which physical ports, and which interface names corresponded to what. br-lan turned out to be LAN/WLAN interface, eth0 was the 1-4 ethernet ports, eth1 was the WAN port, and wlan0 and wlan1 are the two wireless interfaces that are disabled by default.

The port layout on the WDR3700

WAN port is a gigabit port. VLANs can be configured to work on both LAN and WAN.

To create a VLANx on WAN port (eth1) declare an interface as “eth1.x”. It can be further bridged with one of the switch ports if necessary.

For the LAN interface two actions are needed:

Declare “eth0.x” interface
Define switch_vlan configuration for each of the switch ports.
When vlans are defined on the switch ports VLAN0 cannot be used on eth0 any more. So the default router configuration must be changed, otherwise traffic stops on the switch ports and hard reset is needed.

Here is an example. WAN port is trunked on VLAN6 and VLAN4. VLAN6 has dhcp protocol defined. VLAN4 is bridged with one of the switch ports. LAN ports 1-3 are assigned to default VLAN1 (VLAN0 is not possible for some reason). LAN port 4 is assigned to VLAN4 and bridged with WAN.

So all the traffic that goes to switch port 4 will also pass to WAN’s vlan4. Switch port numbering is other way around from physical ports.

Now, the last thing that needs to be dealt with is the fact that unless you are American, BOTH 2.4GHz and 5GHz will likely not be fully available on OpenWrt (you will experience loss of channels 12, 13 and 14 if these are available where you are, as well as potentially incorrect 5GHz channel availability. Power output levels may be incorrect for your country as well).

To sum up the background for this:

• Netgear ships the WNDR3700 with EEPROM chip stating that the router should restrict to US settings. Outside America, they ship software that ignores this chip. This is a bad thing as it means that your OpenWrt can’t work properly for the next two reasons.
• In the USA, it is illegal to ship router firmware that can (through incorrect settings) bypass radio regulations.
• Atheros (the wireless chip maker in the WNDR3700) asks open source folks not to ship compiled builds that bypass radio restrictions, even though their hardware vendors do this. This means if you bought hardware from one of the manufacturers (and you’re one of the 95% of the world’s population that doesn’t live in the USA) that doesn’t set the value in the EEPROM correctly, you’re kinda screwed if you want to use OpenWrt…

Luckily, there are ways to get around this. You can either follow the rather complicated guide, or hope that someone has done this for you and published the binary file for your specific router.

There are two ways to delete all the data in a MySQL database table.

TRUNCATE TABLE tablename; This will delete all data in the table very quickly. In MySQL the table is actually dropped and recreated, hence the speed of the query. The number of deleted rows for MyISAM tables returned is zero; for INNODB it returns the actual number deleted.

DELETE FROM tablename; This also deletes all the data in the table, but is not as quick as using the “TRUNCATE TABLE” method. In MySQL >= 4.0 the number of rows deleted is returned; in MySQL 3.23 the number returned is always zero.

Download Here (Grab the Play Store version quickly, 50% off today!)