The specifics of the SIP information retrieval via DNS queries are described in RFC 3263 – Session Initiation Protocol (SIP): Locating SIP Servers. This RFC details the SIP specificities on top of RFC 2782 – A DNS RR for specifying the location of services (DNS SRV).

In our case, the P-CSCF of the visited domains looks at the Request-URI the UE sent in its Register message. It observes this domain is different than its own, so it concludes the UE is in roaming, so it has to locate the home I-CSCF, in order to see if it can serve this subscriber or not.

So, what steps does the visited P take in order to get to the right home I?

The P performs a NAPTR query for the domain specified in the Request-URI;

This NAPTR thinggie comes from Name Authority Pointer and it is actually a type of DNS RR - Resource Record. The reason behind the invention of this horrible RR is the quest to map everything around the interwebs – we’re mapping you over!!! Resistance is futile !!! HA HA HAAA

Now, seriously, they are trying to map resources like services (like printers, LDAP servers or..why not…I-CSCF servers!!!) to a plain domain name. This NAPTR has a specific structure:

- service name

- set of flags

- regexp rule (yes, regular expressions, we all hate them , or at least I surely do so, at least after the Perl learning tentative I had a while back)

- an order value

- a preference

- replacement

and, even more, they can also come chained in multiple records carefully cascaded to make our poor lives even more miserable.

These NAPTR things are standardized by RFC 2915 and RFC 3403. Good luck with that!

Moving on, let’s take a look at how a P-CSCF NAPTR Query may look like:

OPCODE=SQUERY

QNAME=registrar1.home.net, QCLASS=IN, QTYPE=NAPRT

The NAPTR Response would be like this:

OPCODE=SQUERY, RESPONSE, AA

QNAME=registrar1.home.net, QCLASS=IN, QTYPE=NAPTR

registrar1.home.net       0   IN   NAPTR   50 50 “s” “SIP+D2U” “” _sip._udp.registrar1.home.net

0   IN   NAPTR 90 50 “s” “SIP+D2T” “” _sip._tcp.registrar1.home.net

0   IN   NAPTR 100 50 “s” “SIPS+D2T” “” _sips._tcp.registrar1.home.net

UDP is preferred, as the UDP record appears first – order criteria. The “s” means this is a SRV record. What this P needs to do further on is to perform a SRV Query at the address provided in the NAPTR record first ( _sip._udp.registrar1.home.net) in order to get the services supported by this guy (the I-CSCF).

The SRV Query would look something like this:

OPCODE=SQUERY

QNAME=_sip._udp.registrar1.home.net, QCLASS=IN, QTYPE=SRV

And the SRV Response:

OPCODE=SQUERY, RESPONSE, AA

QNAME=_sip._udp.registrar1.home.net, QCLASS=IN, QTYPE=SRV

_sip._udp.registrar1.home.net   0   IN   SRV   1   10   5060   icscf1_p.home.net

0   IN   SRV   1   0   5060   icscf7_p.home.net

icscf1_p.home.net   0   IN   AAA   5555::aba:abb:abc:abd

icscf7_p.home.net   0   IN   AAA   555::dba:dbb:dbc:dbd

Here there are listed all the I-CSCF proxies in the home.net domain, with their Priority and Weight. The best one will be chosen, according to the rules defined in RFC 2782. As the answer also contains the IP address of the I-CSCF, the visited P-CSCF will forward the REGISTER message to this I-CSCF (here icscf1), on port 5060.

Now, let’s analyze the messages from the IMS – SIP signaling. As I was saying, these messages are tunneled between the eNodeB and PGW via GTPv1-U protocol. Then they reach the P-CSCF and are forwarded in the IMS core. This P-CSCF entity, often called simply P, can be located – and usually it is, specially in the roaming scenarios, … located in the visited network.

Before continuing to the description of each of the messages in the IMS registration, let’s take another look at the 4G – IMS architecture, as well as to the registration flow that we describe:

4G-IMS Architecture

4G-IMS RegistrationFlow

16. P-CSCF Discovery

- described in TS 23.228: section 5.1.1 Procedures related to Proxy-CSCF discovery and

section E.1.1.1 GPRS/EPS procedure for P-CSCF discovery

Because the procedure is pretty straight-forward, I will just copy-paste it from the spec above:

Still, before continuing the IMS registration started in http://www.imacandi.net/windancer/2010/08/26/4g-to-ims-call-flow-register-to-ims.html

I would like to show a preview of the Specs describing this fancy IMS thinggie.

[This information is not structured by me, but it was on tech-invite, before they decided to put a price on all that information.]

The General aspects:

The procedure of IP-CAN (Connectivity Access Network) establishment is described in TS 23.203.

Basically, the IP-CAN is the IP address that the User Equipment gets:

- either during Initial Attach

- or at a Dedicated Bearer creation

that connects that UE to the IMS APN. There are multiple ways of getting this IP address, ways varying from DHCP/DHCPv6, PPP and so on. With regards to IMS, the IP-CAN is given by the PGW, following the PCC rules from its local PCRF.

But, before displaying the actual IP-CAN Session Establishment, let’s take a look at the functional entities and the architectures involved.

The possible scenarios when talking about PCC (Policy and Charging Control) functionality, presented in TS 23.203. I have just copied the architecture pictures. The functional entities are described separately in the same TS 23.203 spec – section 6.2. Functional Entities. I will just underline, where the case requires, which entity from this “PCC Reference Architecture” matches which entity on the 4G architecture.

Before, let’s clarify a few things about these 3 funky pictures:

The acronyms:

HPLMN == Home Public Land Mobile Network

VPLMN == Visited Public Land Mobile Network (the user is in roaming)

SPR == Subscription Profile Repository

TS 23.203 – section 6.2.4 – SPR :

The SPR logical entity contains all subscriber/subscription related information needed for subscription-based policies and IP‑CAN bearer level PCC rules by the PCRF. The SPR may be combined with or distributed across other databases in the operator’s network, but those functional elements and their requirements for the SPR are out of scope of this document.

** The SPR is connected to the PCRF via the Sp interface.

AF == Application Function

TS 23.203 – section 6.2.3 – AF:

The Application Function (AF) is an element offering applications that require dynamic policy and/or charging control over the IP‑CAN user plane behaviour. The AF shall communicate with the PCRF to transfer dynamic session information, required for PCRF decisions as well as to receive IP‑CAN specific information and notifications about IP‑CAN bearer level events. One example of an AF is the P‑CSCF of the IM CN subsystem.

** The AF is connected to the PCRF via the Rx interface.

OCS == Online Charging System

- The OCS is described in TS 32.240. In the PCC architecture we are only interested in its component called Service Data Flow Based Credit Control - its main purpose being to perform online credit control functions.

** The OCS is connected to the PCEF via the Gy interface.

PCRF == Policy Control and Charging Rules Function

TS 23.203 – section 6.2.1 PCRF:

The PCRF encompasses policy control decision and flow based charging control functionalities.

The PCRF provides network control regarding the service data flow detection, gating, QoS and flow based charging (except credit management) towards the PCEF.

The PCRF shall apply the security procedures, as required by the operator, before accepting service information from the AF.

…………….

Here we are talking about 2 PCRF entities:

a) there is only one PCRF involved in the first case, when the UE is NOT in roaming; this is the local/home PCRF

b) in the home routed access and local breakout scenarios there is on one hand the V-PCRF (visited) – the PCRF entity from the visited network and on the other hand the H-PCRF (home) – the PCRF entity from the home network

** The PCRF is connected to the:

- AF via the Rx interface

- SPR via the Sp interface

- BBERF via the Gxx interface

- PCEF via the Gx interface

- H-PCRF and V-PCRF are connected via the S9 interface

BBERF == Bearer Binding and Even Reporting Function

TS 23.203 – section 6.2.7 BBERF:

The BBERF includes the following functionalities:

-     Bearer binding.

-     Uplink bearer binding verification.

-     Event reporting to the PCRF.

-     Sending or receiving IP‑CAN-specific parameters, to or from the PCRF.

* Note: As far as I understand, and in order to somehow “map” the names of the “PCC” entities to the names of the “EPC” entities I’ve first learned about, I believe this BBERF role is actually played by the SGW as we know it from the EPC terminology. Please correct me if I’ve got this wrong.

CORRECTION (further reading on TS 23.203): In the GTP-based 3GPP access network the BBERF entity does NOT apply.

** The BBERF is connected to the PCRF via the Gxx interface.

PCEF == Policy and Charging Enforcement Function

TS 23.203 – section 6.2.2 PCEF:

The PCEF encompasses service data flow detection, policy enforcement and flow based charging functionalities.

This functional entity is located at the Gateway (e.g. GGSN in the GPRS case, and PDG in the WLAN case). It provides service data flow detection, user plane traffic handling, triggering control plane session management (where the IP‑CAN permits), QoS handling, and service data flow measurement as well as online and offline charging interactions.

………………

*Note: The EPC entity assuming the role of the PCEF in the PCC Architecture is the PGW.

** The PCEF is connected to the:

- PCRF via the Gx interface

- OCS via the Gy interface

- OFCS via the Gz interface

OFCS == Offline Charging System

TS 23.203 – section 6.2.6 OFCS:

The Offline Charging System is specified in TS 32.240 [3].

There may be several OFCSs in a PLMN. The default OFCS addresses (i.e. the primary address and secondary address) shall be locally pre-configured within the PCEF. OFCS addresses may also be passed once per IP‑CAN session from the PCRF to the PCEF. The addresses provided by the PCRF shall have a higher priority than the pre-configured ones.

** The OFCS is connected to the PCEF via the Gz interface.

[I'll keep the pictures numbering for future referencing.]

Figure 5.1.1 Overall PCC logical architecture (non-roaming)

This architecture describes the simplest case where the User Equipment is located in his home network – in the network of the operator he subscribed to.

Figure 5.1.2 Overl PCC architecture (roaming with home routed access)

As the picture shows, here only the BBERF (which can be an SGW or a SGSN) is located in the visited network. This implies that the local (visited) PCRF is also to be used when locating the UE. The visited PCRF will contact the home PCRF via the S9 interface.

Figure 5.1.3 Overall PCC architecture for roaming with PCEF in visited network (local breakout)

This is the case where basically the entire access network is a visited network as the UE is concerned about. The BBERF (SGW or SGSN) and the PCEF (GGSN or SGW) [at least as far as the 3GPP implementations go...WiMAX guys, please help me out complete this article] are in the visited network. This implies: the use of (at least) the local (visited) PCRF, possibly the use of a local AF and the existence of a local OFCS.

This being said, let’s see how the IP-CAN Session Establishment process takes place – shamelessly copy-pasted from TS 23.203 – section 7.2 IP-CAN Session Establishment – with the consideration that, at least this spec is concise enough when describing these procedures so that I won’t feel the need to add anything else (afaik):

** Careful with the local notes, as in this single picture there are represented the IP-CAN procedures for all the 3 roaming/non-roaming scenarios described above:

This procedure concerns both roaming and non-roaming scenarios. In the roaming case when a Gateway Control Session is used, the V-PCRF should proxy the Gateway Control Session Establishment information between the BBERF in the VPLMN and the H-PCRF over S9 based on PDN-Id and roaming agreements.

For the Local Breakout scenario (Figure 5.1.3) the V-PCRF shall proxy the Indication and Acknowledge of IP‑CAN Session Establishment over S9 between the PCEF in the VPLMN and the H-PCRF

In the non-roaming case (Figure 5.1.1) the V-PCRF is not involved.

1.   The BBERF initiates a Gateway Control Session Establishment procedure as defined in clause 7.7.1 (applicable for cases 2a during initial attach and 2b, as defined in clause 7.1).

2.   The GW(PCEF) receives a request for IP‑CAN Bearer establishment. A PDN Connection Identifier may be included in the request. The GW(PCEF) accepts the request and assigns an IP address for the user.

3.  The PCEF determines that the PCC authorization is required, requests the authorization of allowed service(s) and PCC Rules information. The PCEF includes the following information: UE Identity (e.g. MN NAI), a PDN identifier (e.g. APN), the IP‑CAN type and the IP address(es), if available, the PDN Connection Identifier received for IP‑CAN Bearer establishment and, if available, the default charging method and the IP‑CAN bearer establishment modes supported. The PDN identifier, IP address(es) and UE identity enables identification of the IP‑CAN session. The IP‑CAN Type identifies the type of access from which the IP‑CAN session is established. If the service data flow is tunnelled at the BBERF, the PCEF shall provide information about the mobility protocol tunnelling encapsulation header. The PCEF may also include the Default Bearer QoS and APN-AMBR (applicable for case 1, as defined in clause 7.1). In case 2a the PCEF may also include charging ID information.

4.   If the PCRF does not have the subscriber’s subscription related information, it sends a request to the SPR in order to receive the information related to the IP‑CAN session. The PCRF provides the subscriber ID and, if applicable, the PDN identifier to the SPR. The PCRF may request notifications from the SPR on changes in the subscription information.

5.   The PCRF stores the subscription related information containing the information about the allowed service(s) and PCC Rules information.

6.   The PCRF makes the authorization and policy decision.

7.  The PCRF sends the decision(s) , including the chosen IP‑CAN bearer establishment mode, to the PCEF. The GW(PCEF) enforces the decision. The PCRF may provide the default charging method and may include the following information: the PCC Rules to activate and the Event Triggers to report. The Policy and Charging Rules allow the enforcement of policy associated with the IP‑CAN session. The Event Triggers indicate to the PCEF what events must be reported to the PCRF.

8.   If online charging is applicable, and at least one PCC rule was activated, the PCEF shall activate the online charging session, and provide relevant input information for the OCS decision. Depending on operator configuration PCEF may request credit from OCS for each charging key of the activated PCC rules.

9.   If online charging is applicable the OCS provides the possible credit information to the PCEF and may provide re-authorisation triggers for each of the credits.

In cases 2a and 2b if the OCS provides any re-authorisation trigger, which can not be monitored at the PCEF, the PCEF shall request PCRF to arrange those to be reported by the BBERF via the PCRF.

10. If at least one PCC rule was successfully activated and if online charging is applicable, and credit was not denied by the OCS, the GW(PCEF) acknowledges the IP‑CAN Bearer Establishment Request.

11. If network control applies the GW may initiate the establishment of additional IP-‑CAN bearers. See Annex A and Annex D for details.

12.  If the PCRF in step 7 requested an acknowledgement based on PCC rule operations, the GW(PCEF) sends the IP‑CAN Session  Establishment Acknowledgement to the PCRF in order to inform the PCRF of the activated PCC rules result.

Many more information on this on TS 23.203. Insist on the QoS parameter interaction.

Specially: Annex 4. 3GPP Accesses (GERAN/UTRAN/E-UTRAN) GTP-based EPC

De data aceasta, insa, este despre viata lui Haydn, 15 lectii de aproximativ jumatate de ora. Haydn era super apreciat – desi talentul lui s-a dezvoltat destul de tarziu. Aveam peste 50 de ani, cand englezii, vazand ca sponsorul lui Haydn, printul de Esterhazy, nu-l lasa sa plece de la curtea lui, se gandesc chiar sa-l rapeasca si sa-l aduca in Londra:

By the mid 1780s, most of Haydn’s works were immediately published, then heard across Europe and, for that matter, North America as well. Haydn’s music was embraced in Spain, adored in Italy and France as well as in Germany and Austria. The English in particular worshiped – and that is not too strong a word – worshiped Haydn’s music. Between 1781 and 1787, the English Publishing Firm of William Forester published 129 works by Haydn, of which 82 were symphonies, for which Haydn btw received a beaucoup bucks.

Starting 1783, various English concert societies began inviting Haydn to come over in England to conduct and compose, hang out and get very rich. Unfortunately, and much to Haydn’s unhappiness, he had to decline any and all such offers to travel abroad. There was no chance that Prince Nicholas ” the big shot” – Esterhazy would grant his capel maestro such an extended leave of absence [.....]

[Printul Esterhazy incerca sa-l tina pe Haydn pentru curtea lui, deoarece era clar ca faima lui Haydn l-ar fi determinat sa nu se mai intoarca, nici la monotonia din palatul Esterhazy, dar nici la sotia lui cu care nu se intelegea deloc.]

Haydn’s popularity in England during the late 1780s was such that the press saw his position as little better than imprisonment. Even so far as to suggest that Haydn BE KIDNAPPED and brought to London. I kid you not. The following appeared in the London Review: “There is something very distressing to the liberal mind in the history of Haydn. This wonderful man, who is the Shakespeare of music, and the triumph of the age in which we live is doomed to reside in the court of a miserable German prince, who is once incapable of rewarding him and unworthy of the honor. Haydn, the simplest, as well as the greatest of men, is resigned to his condition and is content to live in a place little better than a dungeon, subject to the dominant spirit of a petty lord and a clamorous spirit of a scolding wife.”

My my – the word DOES get around…..

String Quartet in C Major, Opus 33, nr. 3, “Pasarea”

Click here to view the embedded video.

The normal lines (and arrows) represent the main protocol:

- while in the 4G environment (UE, eNodeB, MME, SGW, PGW): it is eGTP (GTPv2-C) protocol

- while in the IMS environment(P-CSCF, I-CSCF, S-CSCF, HSS): it is plain IP or Diameter protocol

The dotted lines (and arrows) represent the inner IP protocol messages, which are encapsulated in GTPv1-U header.

The specs impacted by this are:

TS 23.401 : for the 4G UE equipment Initial Attach to the network

TS 23.203 : description of the IP-CAN Session Establishment procedures

TS 23.228 : description of the P-CSCF Discovery procedures

TS 21.905 : vocabulary for 3GPP Specifications

TS 29.061 : interworking of 4G and IMS

TS 29.212 : PCC (Policy and Charging Control) over Gx interface

TS 29.213 : PPC signaling flows and QoS parameters

TS 24.229 : IP multimedia call control over SIP and SDP

RFC 3261 : SIP – Session Initiation Protocol

The IMS protocols are just too many and diverse to list here all the TSs and RFCs related to them.

Let’s describe – shortly, very shortly – the messages exchanged here:

[Some of the message descriptions are simply and shamelessly copy-pasted from the spec :p  ]

1. Attach Request [TS 23.401]

2. Attach Request [TS 23.401]

The eNodeB derives the MME from the RRC parameters carrying the old GUMMEI and the indicated Selected Network. If that MME is not associated with the eNodeB or the old GUMMEI is not available, the eNodeB selects an MME as described in clause 4.3.8.3 on “MME selection function”. The eNodeB forwards the Attach Request message to the new MME contained in a S1-MME control message (Initial UE message) together with the Selected Network and TAI+ECGI of the cell from where it received the message to the new MME.

3. Create Session Request [TS 23.401]

If a subscribed PDN address is allocated for the UE for this APN, the PDN subscription context contains the UE’s IPv4 address and/or the IPv6 prefix and optionally the PDN GW identity. In case the PDN subscription context contains a subscribed IPv4 address and/or IPv6 prefix, the MME indicates it in the PDN address. For Attach Type indicating “Initial Attach”, if the UE does not provide an APN, the MME shall use the PDN GW corresponding to the default APN for default bearer activation. If the UE provides an APN, this APN shall be employed for default bearer activation. For Attach type indicating “Handover”, if the UE provides an APN, the MME shall use the PDN GW corresponding to the provided APN for default bearer activation, If the UE does not provide an APN, and the subscription context from HSS contains a PDN GW identity corresponding to the default APN, the MME shall use the PDN GW corresponding to the default APN for default bearer activation. The case where the Attach type indicates “Handover” and the UE does not provide an APN, and the subscription context from HSS does not contain a PDN GW identity corresponding to the default APN constitutes an error case. If the attach type indicates “Initial Attach” and the selected PDN subscription context contains no PDN GW identity the new MME selects a PDN GW as described in clause 4.3.8.1 on PDN GW selection function (3GPP accesses). If the PDN subscription context contains a dynamically allocated PDN GW identity and the Attach Type does not indicate “Handover” the MME may select a new PDN GW as described in clause PDN GW selection function, e.g. to allocate a PDN GW that allows for more efficient routing. The new MME selects a Serving GW as described in clause 4.3.8.2 on Serving GW selection function and allocates an EPS Bearer Identity for the Default Bearer associated with the UE. Then it sends a Create Session Request (IMSI, MSISDN, MME TEID for control plane, PDN GW address, PDN Address, APN, RAT type, Default EPS Bearer QoS, PDN Type, APN-AMBR, EPS Bearer Identity, Protocol Configuration Options, Handover Indication, ME Identity, User Location Information (ECGI), MS Info Change Reporting support indication, Selection Mode, Charging Characteristics, Trace Reference, Trace Type, Trigger Id, OMC Identity, Maximum APN Restriction, Dual Address Bearer Flag, the Protocol Type over S5/S8) message to the selected Serving GW.

The RAT type is provided in this message for the later PCC decision. The subscribed APN AMBR for the APN is also provided in this message. The MSISDN is included if provided in the subscription data from the HSS. Handover Indication is included if the Attach type indicates handover. Selection Mode indicates whether a subscribed APN was selected, or a non-subscribed APN sent by the UE was selected.. Charging Characteristics indicates which kind of charging the bearer context is liable for. The MME may change the requested PDN type according to the subscription data for this APN as described in clause 5.3.1.1. The MME shall set the Dual Address Bearer Flag when the PDN type is set to IPv4v6 and all SGSNs which the UE may be handed over to are Release 8 or above supporting dual addressing, which is determined based on node pre-configuration by the operator. The Protocol Type over S5/S8 is provided to Serving GW which protocol should be used over S5/S8 interface.

The charging characteristics for the PS subscription and individually subscribed APNs as well as the way of handling Charging Characteristics and whether to send them or not to the P GW is defined in TS 32.251 [44]. The MME shall include Trace Reference, Trace Type, Trigger Id, and OMC Identity if S GW and/or P GW trace is activated. The MME shall copy Trace Reference, Trace Type, and OMC Identity from the trace information received from the HLR or OMC.

The Maximum APN Restriction denotes the most stringent restriction as required by any already active bearer context. If there are no already active bearer contexts, this value is set to the least restrictive type (see clause 15.4 of TS 23.060 [7]). If the P GW receives the Maximum APN Restriction, then the P GW shall check if the Maximum APN Restriction value does not conflict with the APN Restriction value associated with this bearer context request. If there is no conflict the request shall be allowed, otherwise the request shall be rejected with sending an appropriate error cause to the UE.

NOTE 7: The Dual Address Bearer Flag is not used when the Protocol Type over S5/S8 is PMIP.

After all, it is not that complicated in the first place:

[TS 23.228]

TS 23.401, sections 5.3.1 – IP Address Allocation, 5.3.2 – Attach Procedure

It seems that:

1. an UE can simultaneously connect to multiple APNs;

2. an UE can have multiple default bearers per APN connection: for example, one for IPv4 and one for IPv6;

2.a) 2 default bearers per APN connection are possible when the MME does NOT set the Dual Address Bearer Flag; this way, the MME forces the sending of separate IPv4 and IPv6 requests for PDN connectivity;

2.b) if the MME sets the Dual Address Bearer Flag, then it can send a request with dual-stack IPv4v6 and the APN can provide both of these IP addresses at once – this means that there are 2 IP addresses (one IPv4 and one IPv6 ONLY one of each type !) for a SINGLE default bearer;

3. Allocation of these IP addresses to the UE can happen from a local PGW pool or from the PDN. In the later case, the Create Session Response message sent from the PGW to the SGW (and further on to the MME) has PAA = 0.0.0.0, following the later completion of this address;

3.a) If the PGW has nothing to do with the further negotiation of the IP addresses, we are talking about a direct transparent access IP allocation; the PGW is just a proxy;

3.b) If the PGW is actively (protocol dependent) implicated in the IP address allocation, then we are talking about a non transparent access; the PGW is an active part in the IP negotiation: for instance, it may act as a DHCP server for the UE (via SGW, of course) and in the same time as DHCP client – when talking to the APN’s actual DHCP server;

*Note: the role of the PGW in the DHCP allocation case is different from the role of its 3G homologous, the GGSN – this entity playing the role of a DHCP relay agent in this scenario;

*Note: We are talking about the so-called IP-CAN (Connectivity to Access Network) session establishment – which, as far as I understand from TS 29.061, refers to the process of allocating an IP address (IPv4, IPv6 or IPv4v6) by a process other than gathering it from the PGW pool, for instance: via DHCP/DHCP-PD, PPP, IMS CN IM process…etc…

4. IP-CAN can be established at:

a) Initial Attach (default bearer activation) to the APN (in EPC) – Primary PDP Context Activation to the APN (in 3G)

b) after the initial attach, via a dedicated bearer/secondary PDP context

5. The IP assignment can take place:

a) either at the subscription phase – in which case we are talking about a static address

b) or at the IP-CAN session establishment – in which case we are talking about a dynamic address

*Note: Usually, as part of the IP-CAN negotiation ( no matter if this takes place at initial attach of afterwards), the PGW may request the UE to authenticate to the external APN’s AAA server

Ciorba super buna si consistenta, dar tiranul a insistat sa luam felul 2 complet, de 2 persoane; impreuna abia am mancat jumatate. Iar eu a mai trebuit sa tin loc si pentru piureul de castane super genial pe care-l servesc cei de la Excalibur. Acum stam amandoi ca niste gansaci umflati in pat si bem vin sa ajutam digestia

/me happy